Radiation Standards: Scientific Basis Inconclusive, and EPA and NRC
Disagreement Continues (Letter Report, 06/30/2000, GAO/RCED-00-152).

Pursuant to a congressional request, GAO examined the Environmental
Protection Agency's (EPA) and the Nuclear Regulatory Commission's (NRC)
radiation standards, focusing on: (1) whether the U.S. radiation
protection standards have a well-verified scientific basis; (2) whether
federal agencies have come closer to agreeing on standards since GAO
reported on this issue in 1994; and (3) how implementing these standards
may affect the costs of nuclear waste cleanup and disposal activities.

GAO noted that: (1) U.S. regulatory standards to protect the public from
the potential health risks of nuclear radiation lack a conclusively
verified scientific basis, according to a consensus of recognized
scientists; (2) scientists have assumed that even the smallest radiation
exposure carries a risk; (3) this assumption extrapolates
better-verified high-level radiation effect to lower, less well-verified
levels and is the preferred theoretical basis for the U.S. radiation
standards; (4) some say that the model is overly conservative and that
below certain exposure levels, there is no risk of cancer from
radiation; (5) others say that the model may underestimate the risk; (6)
interest among scientists in obtaining a more conclusive understanding
of the effects of low-level radiation has been evident in recent
federally funded initiatives, including a reassessment by the National
Academy of Sciences of the latest research evidence on the risks of
low-level radiation; (7) also, a 10-year DOE research program, begun in
fiscal year 1999, has been specifically addressing the effects of
low-level radiation within human cells, in part to help verify or
disprove the linear model; (8) although GAO recommended as far back as
1994 that EPA and NRC take the lead in pursuing an interagency consensus
on acceptable radiation risks to the public, they continue to disagree
on two major regulatory applications: (a) the proposed disposal of
high-level nuclear waste in a repository at Yucca Mountain; and (b) the
cleanup and decommissioning of nuclear facilities; (9) centrally at
issue between the two agencies is groundwater protection; (10) EPA
applies community drinking water limits for radioactive substances to
groundwater at nuclear sites while NRC includes groundwater and other
potential contamination sources under a less restrictive limit of 25
millirem a year for all means of exposure, an approach that conforms to
internationally recommended radiation protection guidance; (11) as
applied in proposed standards for nuclear waste disposal at Yucca
Mountain, EPA's groundwater approach has been criticized as technically
unsupported by the National Academy of Sciences, which Congress mandated
to recommend standards for the repository; (12) the costs of
implementing different radiation standards vary, depending on the
standards' restrictiveness; and (13) comprehensive estimates of overall
costs to comply with current and prospective standards were unavailable.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  RCED-00-152
     TITLE:  Radiation Standards: Scientific Basis Inconclusive, and
	     EPA and NRC Disagreement Continues
      DATE:  06/30/2000
   SUBJECT:  Nuclear waste management
	     Interagency relations
	     Health hazards
	     Radiation exposure hazards
	     Radioactive waste disposal
	     Safety standards
	     Cost analysis
	     Public health research
	     Environmental monitoring
	     Safety regulation
IDENTIFIER:  DOE Yucca Mountain Project (NV)
	     Superfund Program

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************

GAO/RCED-00-152

Appendix I: Scope and Methodology

32

Appendix II: Major U.S. Radiation Standards

34

Appendix III: Examples of Different Models of Low-Level
Radiation Effects

35

Appendix IV: Overview of Epidemiological Research on
Low-Level Radiation Effects

36

Appendix V: Costs of Different Radiation Standards

40

Appendix VI: Comments From the Environmental
Protection Agency

46

Appendix VII: Comments From the Nuclear Regulatory
Commission

56

Appendix VIII: Comments From the Department of Energy

58

Table 1: Estimated Costs to Achieve Different Soil Cleanup
Levels at Selected DOE Sites and Generic
NRC-Licensed Sites 27

Table 2: Potential Costs to Achieve Different Soil Cleanup
Levels--DOE's, NRC's, and EPA's Analyses 41

Figure 1: The Linear, No-Threshold Model of Low-Level
Radiation Effects 11

Figure 2: Four Models of Low-Level Radiation Effects 35

Figure 3: Cleanup Costs as a Function of Cleanup
Levels--Hazardous Waste Facility, Brookhaven
National Laboratory, 1998 43

DOE Department of Energy

EPA Environmental Protection Agency

GAO General Accounting Office

NRC Nuclear Regulatory Commission

Resources, Community, and
Economic Development Division

B-284378

June 30, 2000

The Honorable Pete Domenici
United States Senate

Dear Senator Domenici:

As the cold war came to a close, the United States shifted its focus from
producing nuclear weapons to cleaning up its nuclear weapons production
facilities. The Department of Energy (DOE), which manages the U.S. nuclear
weapons program, is now cleaning up over a dozen major weapons production
sites around the country. In addition, the nation's nuclear power industry
is starting to decommission over 100 commercial nuclear power plants located
in 31 states, a task that will continue during the coming decades.
Furthermore, DOE is determining the feasibility of constructing an
underground repository to provide for permanently disposing of much of the
nation's highly radioactive waste at Yucca Mountain, Nevada. Until a
repository is operational, federal facilities and nuclear power plants
across the country will continue to store their highly radioactive waste
on-site.

What standards should be used to protect the public from the risks of
exposure to low-level radiation remaining at these sites after the nuclear
materials and wastes have been removed--or, in the case of Yucca Mountain,
to protect the public from exposure to the buried waste--is a question for
which two federal agencies share primary responsibility. The Environmental
Protection Agency (EPA) issues generally applicable public radiation
protection standards and administers the Comprehensive Environmental
Response, Compensation, and Liability Act of 1980 (Superfund), which governs
cleanups of federal and nonfederal facilities. The Nuclear Regulatory
Commission (NRC) issues implementing radiation protection standards as part
of its mandate to regulate civilian sources of nuclear radiation, and it
oversees the decommissioning of commercial nuclear facilities. The states
may also be involved in radiation protection efforts under agreements with
NRC for nuclear facilities within their jurisdictions. Finally, the National
Academy of Sciences has a congressionally mandated role in recommending
radiation protection standards for the Yucca Mountain repository.

Historically, EPA and NRC have sometimes differed over how restrictive U.S.
radiation protection standards should be, as we reported in 1994.1 These
differences have implications for the pace and cost of federal facility
cleanups and commercial decommissioning efforts, as well as for the design
and potential development of the Yucca Mountain repository. Concerned about
these issues, you asked us to examine the scientific basis for the agencies'
radiation protection standards and the costs of implementing them. As agreed
with your office, this report examines (1) whether the current U.S.
radiation protection standards have a well-verified scientific basis, (2)
whether federal agencies have come closer to agreeing on standards since we
reported on this issue in 1994, and (3) how implementing these standards may
affect the costs of nuclear waste cleanup and disposal activities. During
our review, we examined many scientific studies and obtained the views of
recognized scientists on the scientific basis of radiation standards. We
focused mainly on differences in standards for Yucca Mountain and nuclear
cleanup and decommissioning sites because they are prominent current
examples of the debate about standards. In addition, the report includes a
review, performed by a recognized expert in environmental radiation, of
scientific research correlating naturally occurring (background) radiation
levels in the United States and around the world with local cancer rates.
This review was designed to determine whether the research results might
have implications for setting radiation protection standards. (See app. I
for a detailed discussion of our scope and methodology.)

U.S. regulatory standards to protect the public from the potential health
risks of nuclear radiation lack a conclusively verified scientific basis,
according to a consensus of recognized scientists. In the absence of more
conclusive data, scientists have assumed that even the smallest radiation
exposure carries a risk. This assumption (called the "linear, no-threshold
hypothesis" or model) extrapolates better-verified high-level radiation
effects to lower, less well-verified levels and is the preferred theoretical
basis for the current U.S. radiation standards. However, this assumption is
controversial among many scientists. Some say that the model is overly
conservative and that below certain exposure levels, there is no risk of
cancer from radiation. Others say that the model may underestimate the risk.
The research evidence is especially lacking at regulated public exposure
levels--levels of 100 millirem a year and below from human-generated
sources. Interest among scientists in obtaining a more conclusive
understanding of the effects of low-level radiation has been evident in
recent federally funded initiatives, including a reassessment by the
National Academy of Sciences of the latest research evidence on the risks of
low-level radiation, begun in the summer of 1998 and planned to conclude in
2001. Also, a 10-year DOE research program, begun in fiscal year 1999, has
been specifically addressing the effects of low-level radiation within human
cells, in part to help verify or disprove the linear model.

Lacking conclusive evidence of low-level radiation effects, U.S. regulators
have in recent years set sometimes differing exposure limits. In particular,
EPA and NRC have disagreed on exposure limits. Although we recommended as
far back as 1994 that the two agencies take the lead in pursuing an
interagency consensus on acceptable radiation risks to the public, they
continue to disagree on two major regulatory applications: (1) the proposed
disposal of high-level nuclear waste in a repository at Yucca Mountain and
(2) the cleanup and decommissioning of nuclear facilities. Centrally at
issue between the two agencies is groundwater protection. On the one hand,
EPA applies community drinking water limits for radioactive substances to
groundwater at nuclear sites, as a matter of water resource protection
policy. Some of these limits are equivalent to fractions of a millirem a
year. On the other hand, NRC includes groundwater and other potential
contamination sources under a less restrictive limit of 25 millirem a year
for all means of exposure,2 an approach that conforms to internationally
recommended radiation protection guidance. As applied in proposed standards
for nuclear waste disposal at Yucca Mountain, EPA's groundwater approach has
been criticized as technically unsupported by the National Academy of
Sciences, which the Congress mandated to recommend standards for the
repository. However, the Academy recognizes that EPA has the authority to
establish a separate groundwater limit for Yucca Mountain, and EPA believes
its groundwater protection approach for the repository to be technically
justified. As applied to nuclear cleanup and decommissioning sites where
both EPA and NRC may have jurisdiction, the two agencies' different
regulatory approaches have sometimes raised questions of inefficient,
conflicting, dual regulation. There has been little progress in finalizing a
memorandum of understanding, encouraged by the House Appropriations
Committee in August 1999, to resolve EPA's and NRC's conflict about cleanup
standards. Given their historical differences, EPA and NRC may not easily
agree on groundwater protection standards for Yucca Mountain or on their
respective regulatory roles relating to nuclear cleanup and decommissioning
sites. This report contains a matter for congressional consideration
suggesting that, in such a situation, the Congress may wish to help resolve
the agencies' disagreement.

The costs of implementing different radiation standards vary, depending on
the standards' restrictiveness. Generally, the costs increase as the
standards become more restrictive. Comprehensive estimates of overall costs
to comply with current and prospective standards were unavailable, but these
costs could be immense, considering that federal agencies expect to fund
hundreds of billions of dollars in nuclear waste disposal and cleanup
projects over many years in the future. According to DOE's and NRC's
analyses of cleanup options for individual sites, costs per site can be many
millions of dollars higher to comply with more restrictive standards than
less restrictive standards, as might be expected. For example, a 1995 DOE
analysis of cleanup options for plutonium-contaminated test ranges at the
Nevada Test Site estimated $35 million in costs to achieve a
100-millirem-a-year-level, over three times as much to achieve a
25-millirem-a-year level, and over six times as much to achieve a
15-millirem-a-year level. Finally, the analysis showed costs that were over
28 times higher to achieve a 5-millirem-a-year level, illustrating that
compliance costs accelerate rapidly to achieve the most restrictive
protection levels.

We presented a draft of this report to NRC, DOE, and EPA for comment. NRC
found the report to be fundamentally sound, and DOE found it to be factual
and balanced. EPA disagreed with the report's conclusions, particularly our
conclusion that there has been little progress on the finalization of a
memorandum of understanding to resolve EPA's and NRC's conflict about
cleanup standards. However, although the two agencies are developing such a
memorandum, they have had long-standing differences, and we question whether
their latest efforts will resolve these differences without congressional
intervention. All three agencies provided technical comments on the report.
In response to the comments received, we made some changes to the report's
presentation.

Nuclear radiation can be generally categorized as either low-level or
high-level radiation. The low-level range includes exposures up to about
10,000 total millirem, 3 although the term commonly is used to refer to
exposures of a few hundred millirem or less.4 The lower portion of the
low-level range includes natural background radiation levels, and the lowest
portion of this range includes public exposure levels regulated under
various U.S. radiation standards, as shown in appendix II.5 Natural
background radiation levels vary around the world, from below 100 millirem
of exposure a year in some places to several hundred millirem a year in
others, with even higher levels recorded in "hot spots." In the United
States, average natural background radiation exposure is about 300 millirem
a year. In addition, medical practices, such as X-rays and nuclear medicine,
and industrial nuclear operations contribute average public exposures of
about 50 millirem a year and 0.1 millirem a year, respectively. Radiation
from within one's own body, largely from naturally present radioactive
potassium, contributes almost 40 millirem a year, on average. As shown in
appendix II, regulatory public exposure limits vary from a few millirem a
year up to 100 millirem a year. At these levels, radiation is only one of
many environmental and biological events (such as heat) that may alter
(mutate) cell structure, and low-level radiation is commonly considered to
be a relatively weak source of cancer risk.6 To counter these cellular-level
mutations, the human body has active repair processes, although these
processes are not entirely error-free, and their relevance to human cancer
risk remains unclear. Should a radiation-caused cancer develop in one or
more cells, the process may take years, and the source of the cancer will be
verifiable only in exceptional cases, given the current limited
understanding of how cancer develops. Although nearly one in four persons in
the United States dies of cancer from all causes, low-level radiation
presumably accounts for a very small fraction of these cancers, if any.
However, the fraction cannot be quantified.

Federal agencies, and in some instances states, administer U.S. radiation
standards. EPA and NRC administer the majority of the federal standards. EPA
issues environmental radiation protection standards as mandated under
Presidential Reorganization Plan No. 3 of 1970. NRC issues standards as part
of its mandate to regulate civilian sources of nuclear radiation, under the
Atomic Energy Act. (Under the same act, DOE has issued public and worker
exposure limits applicable on-site at the agency's nuclear installations.)
Both EPA and NRC have regulatory roles related to nuclear waste disposal and
nuclear site cleanup and decommissioning. For example, under the Energy
Policy Act of 1992, both have roles at the proposed Yucca Mountain
repository in southern Nevada. The proposed function of the repository is to
receive and dispose of high-level waste from DOE sites and commercial power
plants around the country. EPA has the role of issuing standards to protect
the public from releases of radioactive materials from the facility, and NRC
has the role of issuing technical requirements and criteria and licensing
the facility. Under the act, exposure limits in NRC's final technical
requirements and criteria are to be consistent with the limits in EPA's
final public protection standards. DOE's role at Yucca Mountain will be as
the developer and prospective operator of the repository, and the Department
is pursuing the goal of deciding in 2001 whether to recommend the site to
the President as suitable for nuclear waste disposal. Also involved in Yucca
Mountain oversight are expert advisory bodies, including the Nuclear Waste
Technical Review Board and the National Academy of Sciences, which was
mandated under the Energy Policy Act of 1992 to recommend standards for the
repository. In regard to nuclear cleanup and decommissioning activities,
both EPA and NRC have mandated roles: EPA administers Superfund, the
legislation that governs cleanups of federal and nonfederal facilities, and
NRC regulates the

decommissioning of over 100 active commercial nuclear power plants, as well
as other commercial nuclear facilities, under the Atomic Energy Act.7, 8

Our September 1994 report on radiation protection issues found that U.S.
radiation standards reflected a lack of federal agency consensus on
acceptable radiation risk to the public, as well as a lack of interagency
coordination on standards. Among the reasons we found for the lack of
consensus were differences in agencies' historical missions and legislative
mandates, as well as differences in agencies' regulatory strategies,
particularly in those of EPA and NRC. For example, EPA has historically in
many cases implemented a risk-based radiation protection approach, under
which the agency addresses individual contamination sources, coregulates
chemicals and radioactive substances, and protects both human health and
environmental resources.9 In accordance with its tradition of regulating
chemicals, EPA has generally set a risk of 1 in a million that an individual
will develop cancer in a lifetime as a goal for remediation and has
considered a risk of greater than 1 in 10,000 to be potentially excessive.
EPA's approach has been described as "bottom up," setting a relatively
restrictive risk goal to be pursued through the best available
technology--but allowing less restrictive limits in site-specific
situations. In contrast, NRC favors a dose-based, radiation-specific
protection approach that focuses on human health protection.10 NRC's
protection strategy has been described as a "top down" approach. Compared
with EPA, NRC sets a relatively less restrictive dose limit but reduces
doses (and risks) well below the limit in site-specific situations where the
reductions are "reasonably achievable."11

Scientific Basis

The standards administered by EPA and NRC to protect the public from
low-level radiation exposure do not have a conclusive scientific basis,
despite decades of research. These standards are based on a hypothetical
model of low-level radiation effects. The model, derived from studies of
large populations (in the tens of thousands) exposed to radiation,
extrapolates better-verified high-level radiation effects to lower, less
well-verified levels. The standards protect at annual millirem levels
considerably lower than the better-verified levels. According to a consensus
of scientists, there is a lack of conclusive evidence of low-level radiation
effects below total exposures of about 5,000 to 10,000 millirem. The model
under which these effects are assumed, lacking conclusive evidence, is
called the "linear, no-threshold" hypothesis or model. According to this
model, even the smallest radiation exposure carries a quantifiable cancer
risk. The model, which has been endorsed by national and international
radiation protection organizations and used for many years as a preferred
model in regulating low-level radiation, is a fundamental basis for U.S.
radiation standards. There is interest among scientists in obtaining more
conclusive evidence of radiation effects, and a DOE research program that is
examining cellular low-level radiation effects may eventually help to
develop a better understanding of the cellular processes of radiation cancer
causation.

Conclusive evidence of radiation effects is lacking below a total of about
5,000 to 10,000 millirem, according to the scientific literature we examined
and a consensus of scientists whose views we obtained. At these

levels, authoritative bodies have estimated radiation risks through complex
modeling of the available data,12 and regulators have assumed that even the
smallest radiation exposure carries a risk. This assumption is commonly
referred to as the linear, no-threshold hypothesis or model. Extrapolated
mainly from high-dose effects reported for Hiroshima and Nagasaki survivors,
the linear model assumes that radiation health effects are proportional to
exposure. As figure 1 shows, the model uses a straight line to extrapolate
risks all the way down to zero. From zero upward, the model assumes that as
exposure doubles, risk doubles, and that no entirely risk-free exposure
level or threshold exists.

Radiation protection organizations such as the National Council on Radiation
Protection and Measurements and the International Commission on Radiological
Protection have historically endorsed the model, and U.S. regulators have
used it as a preferred, plausible model, but it is controversial. On the one
hand, the model is widely considered to be a useful, relatively
mathematically simple working hypothesis that may be conservative--that is,
it may not underestimate risks. Regulators make use of the model in doing
risk assessments, regulatory impact analyses, cost-benefit analyses, and
other studies to support their decision-making. In using the model, they are
able to estimate risk reductions and hypothetical lives saved from
regulating at a given exposure level. On the other hand, many scientists
question the validity of the model. The consensus view we encountered is
that the research data on low-level radiation effects are inadequate either
to establish a safety threshold or to exclude the possibility of no effects.
Scientists we contacted and scientific literature we examined generally did
not indicate that any one model clearly best fit the overall data. Instead,
there was evidence that any of several models may "fit." (See app. III.)
Some researchers also said low-level radiation effects are likely too
complicated and variable to be expressed in a single model. There is
evidence that the relationship may vary in individuals, and with the type of
radiation, type of cancer, body organs exposed, sex, and/or age at exposure.
We also found considerable agreement among regulators and scientists that
the linear model may be a conservative "fit" to the data, unlikely to
underestimate risks. However, some said the data support the existence of a
safety threshold below which there are no risks, and others said low levels
of radiation can be beneficial to health. This is a highly controversial
theory, called hormesis, which is in part based on the documented ability of
the body to repair cell damage--referred to as adaptive response. However,
other scientists pointed to studies indicating the linear model may actually
understate radiation risks, especially to fetuses and children.13

Authoritative radiation protection organizations and committees have given
the linear model a qualified endorsement. For example, in 1990, a committee
of the National Academy of Sciences, the Biological Effects of Ionizing
Radiation Committee (BEIR V), reported that the linear model was not
inconsistent with the available research data. According to the committee's
report, at low radiation exposures, risks either less or greater than
linearity--and the existence of a threshold in the low-level dose
range--cannot be excluded, and "the possibility that there may be no risks
from exposures comparable to external natural background radiation cannot be
ruled out. At such low doses, it must be acknowledged that the lower limit
of the range of uncertainty in the risk estimates extends to zero." In
addition, a 1994 report of the United Nations Scientific Committee on the
Effects of Atomic Radiation (UNSCEAR) stated that "there are theoretical
reasons based solely on the nature of DNA damage and repair to expect that
cancer can occur at the lowest doses without a threshold in the response,
although this effect would perhaps not be statistically demonstrable."14
Despite the linear model's unproven and controversial status among
scientists, some scientists said the model is so well accepted that it could
only be superseded on the basis of overwhelming contrary evidence.

Two types of important research into low-level radiation effects have been
conducted. One type of study painstakingly follows the long-term health of
individuals in large populations exposed to radiation, seeking statistically
significant patterns of elevated cancer risks from the radiation exposures.
These are called epidemiological studies. Another type of study subjects
animals or tissue or cell cultures to radiation, seeking biological evidence
of radiation effects. These are called radiobiological studies.

Epidemiological studies may never conclusively prove or disprove the linear
model, according to some scientists. Epidemiological studies have been a key
basis for the linear model, including the research evidence accumulated on
over 85,000 Japanese survivors of the Hiroshima and Nagasaki bomb blasts.
The study, conducted by the international Radiation Effects Research
Foundation, has well established the effects of radiation at high exposure
levels, and scientists have extrapolated this relationship to the low-level
radiation range as well--with considerable inherent uncertainty.15 However,
some scientists have questioned the project's results, asserting among other
concerns that the basic estimates of the Hiroshima and Nagasaki doses (and
their neutron component, for example) still need to be reevaluated, even
after decades of effort devoted to determining these doses.

As noted, epidemiological studies require large study populations for the
research results to be statistically powerful. Epidemiologists consider two
types of epidemiological studies--analytic or ecologic. Analytic studies
either compare individuals who have been exposed to radiation to individuals
who have not been exposed and determine if there are subsequent differences
in their health status (cohort studies) or compare individuals who have a
disease to those who do not to determine if there were differences in the
past exposures of the two groups (case control studies). Ecologic studies
rely on regional data on disease and radiation levels, instead of individual
data, and are considered to be less reliable than analytic studies. Analytic
and ecologic studies have attempted to correlate regional natural background
radiation levels with regional cancer rates at locations in the United
States, Europe, Asia, Brazil, Iran, and other places. A premise related to
such studies is that, if the linear model holds, cancer rates should be
higher at locations where natural background radiation levels are
significantly higher. With the help of an expert consultant, we examined 82
studies, which generally found little or no evidence of elevated cancer
risks from high natural background radiation levels. A large number of
studies reported a lack of evidence of cancer risks, some others reported
evidence of slightly elevated risks, and some others reported evidence of
slightly reduced risks. Overall, the studies' results are inconclusive, but
they suggest that at exposure levels of a few hundred millirem a year and
below, the cancer risks from radiation may be either very small or
nonexistent. (See app. IV.)

Radiobiological studies, particularly molecular studies, may eventually
develop more conclusive scientific evidence of low-level radiation effects
than epidemiological studies, according to scientists. Past radiobiological
research has helped to establish, among other evidence, the genetic effects
of radiation and its effects on individual body organs. Recently, there has
been interest in research into the cellular processes through which
radiation causes cancer, in part in relation to progress in human genome
research in the 1990s.16 Researchers have been obtaining a better
understanding of specific phenomena such as DNA damage and repair,
chromosomal instability, so-called "bystander" effects on neighboring cells,
and cellular adaptation to exposures. Researchers are looking into such
cellular processes for biological signs (or "biomarkers") of radiation
cancer causation. Several stages are apparent in the development of
radiation-caused cancer: DNA damage, misrepair, cancer initiation, cell
proliferation, and tumor promotion (with subsequent malignant
transformation). To date, the first stage in the process is better
understood than the long-term second stage. Since fiscal year 1999, DOE has
funded a research program targeting the biological effects of low-level
radiation at the cellular level, with total funding of almost $220 million
projected over 10 years. The program is considered unique in that it is
designed specifically to better validate the effects of very low levels of
radiation in areas such as cells' response to radiation damage, thresholds
for low-dose radiation effects, and features distinguishing radiation-caused
cell damage from damage from causes internal to the cell. Many scientists
and regulators we interviewed said this type of research could eventually
help to determine more conclusively the effects of low-level radiation and
their potential link to cancer causation.

In October 1998, the National Academy of Sciences contracted to reassess the
linear model and risk estimates for low-level radiation, at the request of
U.S. regulators, including EPA, NRC, and DOE. The regulators, acting through
the Interagency Steering Committee on Radiation Standards, concluded that
enough research progress had been made in the 1990s to warrant the study.
The Academy last did such an assessment in 1990, called BEIR V. The latest
assessment, called BEIR VII, is to be completed by 2001.17 High expectations
have been set for the BEIR VII committee, reflecting the scientific
controversy surrounding the linear model and low-level radiation effects.
For example, in requesting the effort, EPA, DOE, and NRC asked the committee
to focus on areas the agencies do not believe were emphasized in the
previous BEIR V effort. EPA asked the committee to provide a clear
indication of the weight of evidence for risks at low doses and dose rates
and to carefully assess the sources of any inconsistencies in the results
from different epidemiological studies. DOE asked BEIR VII to consider
epidemiological studies on nuclear workers, and NRC asked the committee to
focus especially on evidence of radiation effects at the lowest portion of
the low-level radiation range, at levels where regulators set radiation
standards, and to consider evidence of hormesis. Also, the committee is
committed to fully assessing all pertinent research data, not just the data
that have been traditionally influential, such as the Hiroshima and Nagasaki
data. Because of its broad focus, the BEIR VII assessment could produce
instructive results, but some agency officials and scientists said the
amount of new research data available might not be sufficient to lead the
committee to either fully validate or disprove the linear model. An EPA
official said he expected the BEIR VII work to support the continued use of
the linear model for regulatory purposes.

Public Protection

We reported in 1994 that federal agencies' radiation standards reflected a
lack of consensus on acceptable risk to the public. Today, this situation
persists, and EPA and NRC, the principal federal radiation standard-setting
agencies, continue to disagree significantly on regulatory approaches and
standards related to groundwater protection. Two major instances of their
disagreement are proposed standards for the prospective Yucca Mountain
high-level-waste repository and standards for the cleanup and
decommissioning of federal and commercial nuclear facilities. For these
applications, EPA favors both (1) a public protection limit of 15 millirem a
year from all radiation sources through all means of exposure--called "all
pathway" protection by specialists--and (2) extra protection of groundwater
resources under sites, at limits originally set for community drinking water
systems, equivalent to 4 millirem a year. Alternatively, NRC favors a single
25-millirem-a-year all-pathway public protection limit, within which
groundwater is a potential pathway.18 This disagreement has complicated
planning for the prospective Yucca Mountain high-level waste repository, on
which a national decision is to be made in 2001, as well as day-to-day
planning for facility decommissioning by commercial nuclear operators
licensed by NRC. In both of these cases, it remains to be seen whether EPA
and NRC can resolve their differences or whether the Congress will need to
intervene.

Standards

The disagreement between EPA and NRC on groundwater protection is reflected
in differences in the radiation standards set by the two agencies but
appears most notably in the debate over proposed draft standards for the
Yucca Mountain, Nevada, high-level-waste repository. Radiation standards are
an important part of the ongoing debate about the future of the planned
facility. Both EPA and NRC issued proposed radiation protection standards
for the repository in 1999, NRC in February and EPA in August. The agencies
differ on proposed all-pathway limits (15 millirem a year versus 25 millirem
a year), and especially on extra groundwater protection. The groundwater
issue at Yucca Mountain relates to differences in the two agencies' overall
resource protection policies, as well as technical details. EPA's approach
reflects its attempt to implement a consistent policy, across various
standards, of protecting groundwater as a national resource, in line with
community drinking water standards established in regulations under the Safe
Drinking Water Act. (According to EPA, the policy is based on preventing
pollution before it occurs. If pollution has occurred, the polluter should
be responsible for the costs of cleanup.) On the other hand, NRC believes
its all-pathway approach is fully protective of human health at Yucca
Mountain and elsewhere. In the Commission's view, EPA's drinking water
standards were not originally intended for an application such as Yucca
Mountain, and the Commission questions EPA's technical basis for proposing
extra groundwater protection.19, 20

NRC has been joined in its views by DOE and some others who have commented
on EPA's proposed Yucca Mountain standards, including the National Academy
of Sciences. The Academy has questioned the technical basis for EPA's extra
groundwater protection approach. Specifically, the Academy, together with
NRC, DOE, and other commenters, has asserted, first, that EPA has not
provided a technical rationale for its approach. By contrast, according to
these commenters, NRC has a technically based rationale for its approach
that is in accord with internationally recommended radiation protection
practices. Second, the Academy and others have pointed out that the drinking
water concentration limits to be applied to groundwater at the repository
are outdated, reflecting doses and risks that are inconsistent. These limits
consist of dozens of numerical maximum contaminant levels for radionuclides,
expressed in picocuries per liter, which reflected consistent doses and
risks when they were

established in regulations implementing the Safe Drinking Water Act of
1976.21 These limits are outdated under the latest risk estimation methods.

In particular, the Academy, in its congressionally-mandated role of
recommending reasonable standards for protecting health and safety at the
repository, has questioned EPA's groundwater protection approach for Yucca
Mountain. The Academy did not propose separate groundwater protection
standards for the repository in its own technical recommendations for the
facility, which were issued in 1995. The Academy's November 1999 comments on
EPA's draft standards directly opposed such an approach, calling it
"scientifically unsupported," adding little or no public health benefit.22
According to the Academy, EPA has the authority to establish a separate
groundwater limit for Yucca Mountain but has not presented a technical
rationale for doing so. In addition, NRC and DOE have commented that EPA has
not done a comprehensive analysis of the health benefits and costs of its
groundwater approach for Yucca Mountain. EPA has issued a draft regulatory
impact analysis to accompany its draft Yucca Mountain standards, in
accordance with Executive Order 12866, which calls for such an analysis if
the regulatory action is significant (for example, raises novel legal or
policy issues). However, the draft regulatory impact analysis was limited in
scope (stating that data were lacking for a fuller discussion), and the
document did not analyze the specific impact of EPA's groundwater protection
approach for the repository.

EPA recognizes that the drinking water contamination limits that are to be
applied at the repository are not scientifically up to date. They are based
on 1970s-era methods of radiation dose estimation, which have been
superseded. The limits were originally intended to be equivalent to 4
millirem a year of exposure. However, under updated dose estimation methods,
they no longer reflect 4 millirem a year. Instead, using a commonly accepted
dose conversion factor, they reflect varying annual millirem levels and
acceptable risks, and some reflect millirem levels up to a thousand times
lower than average U.S. natural background radiation levels. A few of the
limits are equivalent to well above 4 millirem a year, but many are
equivalent to fractions of a millirem a year.23 NRC has commented that
because groundwater is expected to be the dominant exposure pathway at Yucca
Mountain, these limits will be the de facto overall protection standards for
the repository.

According to EPA officials, the agency's proposed standards and groundwater
protection approach for Yucca Mountain are justified on policy grounds and
are technically justifiable as well. The agency has applied drinking water
standards to groundwater at the repository, EPA officials said, because it
desires to protect groundwater as an environmental resource (for drinking
water and irrigation needs) in a region where the population has been
growing quickly. In addition, the agency has a general policy of
coregulating chemicals and radionuclides in groundwater, and EPA officials
said the standards for Yucca Mountain should be in accord with this policy.
EPA officials agreed that the agency has not done a comprehensive analysis
of the health benefits and costs of the agency's groundwater approach for
Yucca Mountain, but they believe their regulatory approach has fully
addressed the pertinent overall technical issues related to setting
radiation protection standards for the site. They said they are developing
an expanded regulatory impact analysis to accompany their final standards,
which will not constitute a specific technical rationale for their extra
groundwater protection approach but will address technical and cost issues
related to the implementation of their standards, as the Academy
recommended.

While recognizing that the drinking water concentration limits to be applied
at Yucca Mountain are out of date, reflecting inconsistent doses and risks,
EPA officials said the agency is in the process of updating the limits and
expects to complete this effort by the fall of 2000. They said that if the
final Yucca Mountain standards are issued before then, the updated limits
will be incorporated into them. The officials noted that under a "no
backsliding" provision of the 1996 Safe Drinking Water Act Amendments, any
updated drinking water standards for radionuclides must maintain equal or
greater levels of public health protection. According to EPA, this provision
precludes the agency from raising any of the concentration limits, even in
attempting to achieve greater uniformity in doses or risks.24 A draft
version of the proposed new limits indicates that EPA may not change any of
the limits.25

As EPA and NRC prepare to issue final standards for Yucca Mountain, it is
not evident that the two agencies and the Academy will agree on appropriate
groundwater protection standards for the repository. If they do not agree,
EPA's final Yucca Mountain standards, expected to be issued in the summer of
2000, will lack a degree of technical consensus that would add to their
credibility and acceptability. Aware of the conflict between EPA and NRC
over standards for Yucca Mountain, the Congress, in March 2000, passed
legislation retaining EPA's standard-setting authority while (1) allowing
the Academy and NRC to report to the Congress any major disagreements they
may have with EPA's final standards and (2) delaying the issuance of final
standards for Yucca Mountain until June 2001. On April 25, 2000, the
President vetoed the bill, in part because it would have limited EPA's
authority to issue radiation standards, and on May 2, 2000, the Senate
upheld the veto.

Site Cleanup and Decommissioning

Both EPA and NRC developed nuclear site cleanup and decommissioning
standards in the 1990s, attempting to facilitate the massive national effort
to clean up nuclear contamination at many federal and commercial nuclear
sites--including DOE's nuclear weapons complex and commercial nuclear power
plants licensed by NRC --that are closing down after decades of operations.
In 1995, EPA drafted cleanup standards reflecting the agency's groundwater
protection approach, which includes 15-millirem-a-year all-pathway
protection, plus separate groundwater protection to drinking water
standards. However, the agency withdrew these standards in 1996, before they
were finalized, after other agencies objected to them. Subsequently, in
1997, EPA implemented the same approach through nonbinding Superfund
guidance. Also in 1997, NRC finalized its own cleanup
standards--decommissioning standards for its licensees--reflecting
25-millirem-a-year all-pathway protection.26 In correspondence with NRC, EPA
has objected to NRC's standard as potentially not protective in all cases.27
In some cases, both EPA's and NRC's approaches have both been applied to the
same site, raising questions about potential dual regulation.

EPA and NRC have long disagreed on standards for cleaning up and
decommissioning the nation's nuclear sites. As far back as 1992, the two
agencies signed a memorandum of understanding, agreeing to avoid unnecessary
duplication of regulatory requirements. In 1994, we reported that EPA and
NRC were involved in potentially costly dual regulation of NRC's licensees.
Our previous report's recommendation that the agencies pursue consensus on
acceptable risk was a factor in the establishment in 1995 of the Interagency
Steering Committee on Radiation Standards, which is cochaired by EPA and
NRC. However, despite numerous staff initiatives and some progress in
cooperation, this committee has not resolved major issues between the two
agencies. The two agencies have continued to use separate approaches in
setting standards for cleaning up and decommissioning nuclear sites,
especially when groundwater protection is involved. Consequently, perceived
dual regulation by EPA and NRC continues to complicate the cleanup and
decommissioning process at some sites where both agencies' standards may
apply, potentially causing duplication of effort and regulatory delays,
adding to facilities' compliance costs, and raising public questions about
what cleanup levels are appropriate and safe.28

For example, in individual situations at NRC-licensed sites, EPA has
indicated that it might not view cleanups performed to NRC's standards as
adequately protective under its Superfund guidance. EPA considers such
potentially conflicting situations to be exceptions, not the rule. EPA
raised the matter of the standards' adequacy twice in 1999 in connection
with nuclear power plants in Maine and Connecticut where the decommissioning
process is under way, and again in connection with the West Valley, New
York, nuclear site, a DOE-operated facility where NRC has a legislatively
mandated regulatory role. In such situations, the licensee may construe
EPA's involvement as a warning that EPA could reevaluate the adequacy of a
cleanup that has met NRC's requirements. In the New England cases, the
licensee has thus far responded to the prospect of dual regulation by
proceeding with its decommissioning plans, assuming that it will be able to
comply with both agencies' standards.29 In the West Valley case, it remains
to be seen whether NRC's or EPA's approach will finally be chosen for the
site.

Our 1994 report found that U.S. radiation standards reflected a lack of
federal agency consensus on acceptable radiation risk to the public,
particularly between EPA and NRC. We recommended that the two agencies take
the lead in pursuing interagency consensus on acceptable radiation risks to
the public. In succeeding years, EPA and NRC have attempted to resolve their
conflict over cleanup standards by means of a memorandum of understanding
that would clarify their potentially conflicting, dual regulation of
NRC-licensed sites. In August 1999, the House Appropriations Committee
strongly encouraged the two agencies to develop such a memorandum and
directed them to report to the Congress by May 1, 2000, on the status of
their efforts to do so.

In early May 2000, EPA and NRC informed the Committee by separate letters
that they are developing such a memorandum, although a jointly drafted
version of the memorandum does not yet exist. To date, EPA officials see
such a memorandum as providing an outline of consultation procedures for EPA
and NRC to use during NRC's decommissioning process when NRC requests EPA's
assistance. It is unclear whether the memorandum will consider whether EPA
should retain its authority to conduct a Superfund evaluation of an
NRC-licensed or -decommissioned facility when a stakeholder requests such an
evaluation. EPA believes it should retain this authority and has provided
guidance to its regional offices on how to proceed when a stakeholder asks
for an evaluation. According to this guidance, EPA believes that at the vast
majority of NRC-licensed sites, cleanups that meet NRC's standards will also
meet Superfund standards. An NRC draft version of the memorandum assumes
that EPA will defer to NRC's radiological cleanup and decommissioning
standards and regulations and will exempt most NRC-regulated sites from the
application of the Safe Drinking Water Act and Superfund. NRC's version
reflects the Commission's view that clearly delineated jurisdictions for the
two agencies are needed.

Long-term costs related to complying with current and prospective U.S.
radiation standards have generally not been comprehensively estimated, but
these costs will be immense, likely in the hundreds of billions of
dollars.30 The potential size of these costs is reflected in DOE's long-term
funding estimates for high-level waste disposal and nuclear cleanup
projects. These estimates total more than $200 billion over many decades and
could increase, according to DOE. In addition, DOE's, NRC's, and EPA's
analyses of individual nuclear sites' cleanup options show that
site-specific compliance costs can vary significantly depending on the
restrictiveness of the standards. As might be expected, the analyses show
that compliance with more restrictive standards--for example, exposure
limits of a few millirem a year--costs considerably more than compliance
with less restrictive standards--for example, limits of 100 millirem a year.
Site-specific DOE analyses estimate multimillion-dollar cost differences
between less restrictive and more restrictive protection levels, in some
cases. These analyses further show faster rising costs to achieve the most
restrictive protection levels.

To comply with high-level waste standards, DOE has projected multibillion
dollar costs, whether or not the Yucca Mountain repository is developed.
Radiation protection standards for Yucca Mountain have not been finalized,31
but DOE has estimated funding of over $43 billion (in 1998 dollars) for the
total repository system to 2116, in large part to help ensure that the
public is protected from exposure to the waste stored there. According to
DOE's latest estimates, long-term funding for the repository could go over
$55 billion. Alternatively, if the repository is not built, DOE has
estimated expenditures of about $52 billion to $57 billion to store
high-level waste for 100 years at existing sites around the country. From
the enactment of the Nuclear Waste Policy Act of 1982 through fiscal year
1998, DOE spent about $7 billion (in 1998 dollars) for its repository
program.32

EPA, NRC, and DOE have not estimated the total difference in compliance
costs under the conflicting standards proposed by EPA and NRC for Yucca
Mountain. DOE officials said that most of the projected increases in the
repository's costs can be associated with design changes resulting from the
combination of EPA's proposed groundwater standards, the need to provide the
level of confidence in the repository's performance required for a rigorous
NRC licensing process,33 and the influence of external oversight advisory
bodies and peer review panels. DOE estimated that cost increases have
totaled over $10 billion since 1993 to achieve added confidence that
restrictive performance and radiation protection requirements can be met
over thousands of years. For example, in 1993 a
repository-performance-related increase of $7 billion (in 1999 dollars) came
from a design change involving the use of more robust, cylindrical waste
containers. Furthermore, DOE is considering an additional cost of $3.7
billion (in 1999 dollars) for water-diverting titanium drip shields (over
the waste containers) intended to protect the repository from potential
water incursion, as well as to meet EPA's proposed groundwater protection
requirements in a rigorous NRC licensing process. Further design
enhancements to achieve a cooler repository, which would keep temperatures
in the repository below the boiling point of water, as favored by the
Nuclear Waste Technical Review Board, could add about $2 billion more to
costs. According to DOE and NRC officials, the latest design is essentially
an attempt to achieve a virtually "no radioactive release" facility at Yucca
Mountain. These officials maintain that such a design may not be needed to
implement less restrictive standards such as 15-, 25- , or
100-millirem-a-year all-pathway exposure limits.

Although comprehensive data on the costs of complying with nuclear site
cleanup and decommissioning standards were unavailable, these costs could be
immense in the long term. For example, complying with EPA Superfund cleanup
and other environmental requirements may cost DOE several billion dollars
annually, judging from the fact that the agency's fiscal year 1999
appropriation for the overall environmental management and cleanup of its
nuclear facilities was about $5.8 billion. The agency's compliance
activities could cost hundreds of billions of dollars and could extend for
decades into the future, according to long-term funding estimates for
environmental cleanup projects. The agency has projected funding for
environmental cleanup at its nuclear sites from fiscal year 2000 through
fiscal year 2070 to be anywhere from about $151 billion to $195 billion (in
1999 dollars). DOE spent about $52 billion for cleanup from fiscal year 1989
through fiscal year 1999. The overall bill for DOE's nuclear cleanup is
uncertain and could go higher. According to DOE, the future costs of many of
its cleanup programs are difficult to quantify because many projects are
still in the planning stage. Moreover, as we noted in 1999, the data on some
sites may not be reliable, in part because many field sites based their cost
estimates on assumed cleanup levels that have not been finalized.34 In
addition, the operators of NRC-licensed nuclear facilities, including over
100 power plants, may spend over $38 billion for decommissioning in coming
decades, according to the Nuclear Energy Institute.

Even though EPA, NRC, and DOE generally did not have estimates for all U.S.
nuclear sites of the costs of complying with different cleanup standards to
achieve different protection levels,35 officials from these agencies said
that achieving more restrictive protection levels, especially the most
restrictive levels--in the range below 100 millirem a year--can be
considerably more expensive. To support their statements, they cited various
generic and site-specific cost analyses conducted by DOE, NRC, and EPA. Our
examination of these cost analyses generally confirmed higher estimated
costs to comply with more restrictive cleanup levels for contaminated soil,
as might be expected. Conversely, cost estimates were considerably less to
comply with less restrictive soil cleanup levels. For example, analyses
comparing the costs of achieving EPA's and NRC's conflicting all-pathway
cleanup levels--15 millirem a year and 25 millirem a year,
respectively--show cost differences in the millions of dollars for some
sites--and even greater cost differences to achieve cleanup levels below 10
millirem a year. Analyses also show potential multimillion-dollar cost
differences between the 15 to 25 millirem-a-year range and the less
restrictive 100-millirem-a-year level.

Among the analyses we examined were a generic analysis by NRC to support its
decommissioning standards and numerous site-specific cost analyses done by
DOE in the course of analyzing cleanup options for its nuclear weapons
complex. The estimated costs of meeting different soil cleanup levels for
selected sites are summarized in table 1 and discussed in more detail in
appendix V.

 Dollars in millions
                        Cost of
 Agency/site/           100         Cost of    Cost of     Cost of less
 analysis date          millirem    25 millirem15 millirem than 10 millirem
                        a year      a year     a year      a year

 DOE/Nevada Test Site                                      1,003a
 and test ranges/1995   35          131        240

 DOE/Brookhaven
 Laboratory waste       15.9        24.4       28.2        64.5b
 facility/1998
 NRC-licensed power
 plantc /1997           0.17        0.31       0.41        1.44d
 NRC-licensed metal
 extraction facilityc   5.30        6.21       7.33        13.86d
 /1997

Note: Totals do not represent overall estimates of cleanup costs, which may
include costs for activities other than soil cleanup, including the
decontamination and removal of equipment and structures, as well as liquid
waste treatment.

a5 millirem a year.

b1 millirem a year. Totals are present values.

cGeneric site.

d3 millirem a year.

Source: GAO's presentation of data from DOE and NRC.

As shown in table 1, for the listed sites, the estimated costs to achieve
different soil cleanup levels vary, from hundreds of thousands of dollars in
some cases to hundreds of million dollars in other cases. The cleanup costs
also accelerate faster to achieve the most restrictive levels. For example,
for the Nevada Test Site, from a 100-millirem-a-year baseline, the costs
increase over three times and over six times to achieve the
25-millirem-a-year and 15-millirem-a-year levels, respectively, but over 28
times to achieve the 5-millirem-a-year level. Similarly, for the Brookhaven
facility, from a 100-millirem-a-year baseline, the costs increase about 53
percent to achieve the 25-millirem-a-year level, about 77 percent to achieve
the 15-millirem-a-year level, but over 300 percent to achieve the
1-millirem-a-year level. DOE and NRC officials said the large cost
variations for different sites reflected site-specific factors, including
the ratio of soil and building contamination, exposure and land-use
scenarios, and waste disposal options. Officials said cost factors include
not only the degree of on-site remediation, but also soil sampling and
analysis to demonstrate compliance with standards, as well as procedural and
other costs.36

Fewer analyses of the costs of complying with EPA's extra groundwater
protection approach for nuclear cleanups were available. However, DOE's
groundwater remediation analyses for aquifers at the Idaho National
Engineering and Environmental Laboratory and at the Brookhaven National
Laboratory showed that the additional costs of achieving drinking water
standards at these sites could approach several hundred million dollars and
a few million dollars, respectively, depending on the remediation option
chosen. (See app. V.)

Although conclusive scientific evidence of the effects of low-level
radiation is lacking and may not soon be found, U.S. regulators still have
the challenge of developing radiation standards that represent their best
estimates of acceptable radiation risks to the public. In this regard, as a
national decision on high-level-waste disposal at Yucca Mountain approaches,
and as EPA and NRC both continue to administer the cleanup and
decommissioning of nuclear sites, it is important that the two agencies
agree on protection approaches and policies for these regulatory
applications. However, it does not appear that EPA and NRC will readily
agree on appropriate groundwater protection approaches for Yucca Mountain.
Also, EPA and NRC have been working on a memorandum of understanding since
before August 1999, when the House Appropriations Committee encouraged them
to clarify their conflicting regulatory roles related to nuclear facility
cleanup and decommissioning, with little progress. Looking back, we note
that they have not successfully addressed this matter since at least 1994,
when we recommended that they pursue consensus on acceptable radiation risks
to the public. Given the agencies' historical differences and lack of recent
progress, without congressional intervention, they may not resolve their
differences.

It is of note that EPA and NRC, while disagreeing over appropriate public
protection levels, are both regulating at levels where the harm of radiation
and the health benefits of radiation standards may not be clearly
demonstrable. Regulating at these levels, well below the range where
radiation effects have been conclusively verified, is essentially a policy
judgment. Such an approach may arguably be prudent, in accordance with
regulatory use of the linear model, which both agencies endorse. However, it
will also be expensive, because compliance costs accelerate to achieve the
lowest exposure levels, as our work confirms. The potential acceptable
risks, health benefits, and costs of EPA's and NRC's differing regulatory
approaches will be of interest to the Congress as it continues to focus on
nuclear health and safety issues of national importance, such as the
proposed Yucca Mountain repository and the cleanup and decommissioning of
federal and commercial nuclear sites. These risks, benefits, and costs will
also affect the public's belief in and acceptance of any resolution of their
conflicting viewpoints that the two agencies may achieve.

The congressional committees of jurisdiction may wish to reconcile EPA's and
NRC's policy differences on groundwater protection for Yucca Mountain. Also,
in connection with the two agencies' efforts to complete a memorandum of
understanding relating to the cleanup and decommissioning of nuclear sites,
these Committees may wish to clarify the agencies' regulatory
responsibilities.

We provided NRC, DOE, and EPA with a draft of this report for their review
and comment. NRC found the report to be fundamentally sound and said it
should help the Congress understand the long-standing differences between
EPA and NRC. NRC supported our conclusions that federal agencies should
agree on decommissioning and high-level waste policies and approaches to
ensure consistent standards and public protection. DOE found the report to
be factual and balanced. EPA disagreed with the report, in separate letters
from its Office of Radiation and Indoor Air and Office of Emergency and
Remedial Response. The Director, Office of Radiation and Indoor Air, said
EPA interprets the information presented in our report differently, and as a
result, EPA disagrees with the report's conclusions. The Director, Office of
Emergency and Remedial Response, said, among other comments, that the report
inaccurately portrays EPA and NRC as making little progress in their
negotiations on a memorandum of understanding to clarify the two agencies'
regulatory roles and responsibilities related to the cleanup and
decommissioning of nuclear facilities. EPA mentioned recent and continuing
efforts by the two agencies to better clarify their respective regulatory
roles through such a memorandum. While recognizing these recent initiatives,
we note that since as long ago as 1992, the two agencies have been
unsuccessful in addressing this matter, and on this basis we still question
whether the two agencies' most recent initiatives will resolve their
differences without congressional intervention.

NRC, DOE, and EPA provided technical clarifications to the draft report,
which we incorporated into the final report where appropriate. EPA's, NRC's,
and DOE's comments and our evaluation of them are included in appendixes VI,
VII, and VIII.

As arranged with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 20 days after
the date of this letter. At that time, we will send copies to the Honorable
Carol Browner, Administrator, Environmental Protection Agency; the Honorable
Richard Meserve, Chairman, Nuclear Regulatory Commission; and the Honorable
Bill Richardson, Secretary of Energy. We will also make copies available to
others upon request.

If you have any questions about this report, please contact me or (Ms.) Gary
L. Jones on (202) 512-3841. Key contributors to this assignment were Duane
G. Fitzgerald and Dave Brack.

Sincerely yours,

Jim Wells, Director, Energy, Resources,
and Science Issues

Scope and Methodology

To conduct our review of U.S. radiation standards, we obtained testimonial
and written documentation from many dozen recognized scientists in the field
of radiation research and radiation protection, including both active
research scientists and representatives on national and international
radiation protection committees; officials of federal agencies principally
responsible for regulating public radiation protection and conducting
scientific research into the health effects of low-level radiation,
including EPA, NRC, DOE, the Department of Defense, and the Department of
Health and Human Services; state radiation protection officials and the
Conference of Radiation Control Program Directors; officials of radiation
protection organizations, such as the National Council on Radiation
Protection and Measurements, the National Academy of Sciences, and the
International Commission on Radiological Protection; environmental and
nuclear industry representatives; and individual radiation researchers in
government, industry, and academia.

To examine the scientific basis of U.S. radiation standards, we obtained
expert views representing various viewpoints in the controversy over
low-level radiation and the linear model. In addition, we hired an expert
consultant to help synthesize data and draw conclusions relating to the
status of studies correlating worldwide natural background radiation levels
to cancer rates, as well as their possible implications for setting
radiation standards. The consultant, Dr. Thomas Gesell, Professor of Health
Physics, Idaho State University, is a recognized expert in environmental
radiation. (Results based on the consultant's work, representing GAO's
views, are presented in detail in app. IV.) Furthermore, we asked several
experts in the radiation protection field to informally review and comment
on the accuracy of a draft version of our report, and we incorporated their
suggested changes where appropriate.

To examine whether federal agencies have come closer to agreeing on
radiation standards since we reported on this matter in 1994, we obtained
views and documentation from agency officials, the Interagency Steering
Committee on Radiation Standards, and the Conference of Radiation Control
Program Directors. We examined various federal and state radiation
standards, as well as the regulatory and compliance activities of various
federal agencies. However, we mainly focused on prominent current instances
of disagreement between EPA and NRC, over radiation protection standards for
high-level waste disposal and for the cleanup and decommissioning of nuclear
facilities.

To examine the costs of radiation standards, although comprehensive
estimates of these costs were unavailable, we obtained indications of these
costs through available agency radiation-related cost and funding data. To
examine the costs of different standards and protection levels, although
comprehensive estimates of these differences were unavailable, we obtained
indications of these costs through examining generic and site-specific
agency cost analyses. We especially looked for analyses that showed
estimated costs to achieve different radiation exposure levels, and we took
steps to reasonably ensure that the analyses represented the best available
data. Specifically, we obtained (1) analyses agencies considered
representative of the best available data, (2) information on internal and
external peer review and other quality controls steps that agencies may have
taken in connection with the analyses, and (3) agency officials' views on
any data limitations in the analyses.

We performed our review between July 1999 and June 2000 in accordance with
generally accepted auditing standards.

Major U.S. Radiation Standards

 Standard/agency                  Numerical limit
 General standards
 General public/NRC (10 C.F.R.
 20)                              100 millirem/year
 Source-specific standards
                                  Radium 226, 228: 5 picocuries/gram
 Uranium mill tailings/EPA, NRC   surface,
 (40 C.F.R. 192; 10 C.F.R. 40,    15 picocuries/gram subsurface
 App. A)                          Radon 222: 20
                                  picocuries/square-meter-seconda
 High-level waste operations/NRC
 (10 C.F.R. 60)                   100 millirem/year
 Spent fuel, high-level waste,
 transuranic waste disposal/EPA   All pathway: 15 millirem/year
 (10 C.F.R. 191)                  Groundwater 4 millirem/yearb
 Yucca Mountain high-level waste
 (proposed)/EPA (64 Fed. Reg.     All pathway: 15 millirem/year
 46976)                           Groundwater 4 millirem/yearb
 Yucca Mountain high-level waste
 (proposed)/NRC (64 Fed. Reg.     25 millirem/year all pathway
 8640)
 Low-level waste/NRC (10 C.F.R.
 61)                              25 millirem/year

 Drinking water/EPA (40 C.F.R.    Radium: 5 picocuries/liter
 141)                             Gross alpha: 5 picocuries/liter
                                  Beta/photon: 4 millirem/yearb
 Uranium fuel cycle/EPA (40
 C.F.R. 190)                      25 millirem/year
 Superfund cleanup/EPA (40 C.F.R. Risk range goals: 1 in 10,000 to 1 in 1
 300)                             millionc
 Decommissioning/NRC (10 C.F.R.
 20)                              25 millirem/year
 Occupational standards
 Occupational Safety and Health
 Administration, NRC, DOE (29
 C.F.R. 1910; 10 C.F.R. 20; 10    5,000 millirem/year
 C.F.R. 835)

aA picocurie is a trillionth of a curie, which is a commonly used unit of
measurement of the activity of radiation.

bRadioactivity from human-made radionuclides in community drinking water
systems.

cLifetime risk of an individual's getting cancer.

Examples of Different Models of Low-Level Radiation Effects

Figure 2 shows different models of low-level radiation effects that could
fit the available research data. The linear model is used and endorsed by
regulators and radiation protection organizations. The threshold model is
preferred by those who interpret the data as showing that there are no
effects below a certain exposure level. The higher-risk model is preferred
by those who interpret the data as showing higher risks than the linear
model. The lower-risk model is preferred by those who interpret the data as
showing lower risks than the linear model.

Overview of Epidemiological Research on Low-Level Radiation Effects

Epidemiological research has been part of the scientific basis for the
linear model and radiation standards. However, epidemiology may not soon
fully verify or disprove low-level radiation effects. Specific
epidemiological research correlating natural background levels in the United
States and around the world with cancer rates has been generally
inconclusive, showing mixed results. Much of this research has used
methodologies that have been widely considered too limited for the research
to be influential in setting radiation standards.

Several U.S. agencies are involved in epidemiological research on low-level
radiation effects, including the Department of Energy (DOE), the Department
of Health and Human Services (HHS), and the Department of Defense (DOD). In
fiscal year 1999, about $41 million in DOE funding went for epidemiological
research, including about $24 million provided to HHS for epidemiological
and environmental research at DOE's nuclear sites, under a 1994 memorandum
of understanding that established independent management of such research.
(In comparison, in fiscal year 1999, about $12 million for radiobiological
research was funded within DOE's Office of Biological and Environmental
Research, and DOD provided an estimated $11 million for radiobiological
research at its Armed Forces Radiobiology Research Institute.)

Epidemiological results have been a key basis for the linear model,
including the study of over 85,000 Japanese survivors of the Hiroshima and
Nagasaki bomb blasts. The United States has for many years participated in
this study, conducted by the international Radiation Effects Research
Foundation, with funding by DOE--$14 million in fiscal year 1999--through
the National Academy of Sciences. The still-continuing study has helped to
establish the effects of radiation at levels above 10,000 millirem, for
which the data show a relationship between exposures and health effects that
is generally consistent with a linear model. With a considerable degree of
inherent uncertainty, scientists have extrapolated this relationship to the
low-level radiation range as well. Based in large part on the Japanese data,
major radiation protection organizations have endorsed the assumption of a
linear dose-response relationship at far lower public exposure levels, down
to those commonly regulated--100 millirem a year and below.

Some scientists have questioned the foundation's work, asserting that the
study understates radiation risk because it necessarily focuses on bomb
survivors, who likely were the healthiest of the blast victims. Others
assert that the important neutron component of the estimated Hiroshima dose
is questionable, calling into doubt the overall Hiroshima-Nagasaki dose
estimates dating from 1986. Recognizing that the Japanese data have
historically been central to radiation risk estimation and that credible
Hiroshima and Nagasaki dose estimates are needed, DOE and the National
Academy of Sciences are conducting a reassessment of the Hiroshima neutron
doses. Results are expected in late 2000. According to DOE, it is likely
that new Radiation Effects Research Foundation dose estimates for Hiroshima
and Nagasaki will be derived and that they will be issued in 2001.

In addition, there have been many epidemiological studies of U.S. and
foreign nuclear workers, medical patients, miners, and others exposed to
various levels of low-level radiation. For example, historically DOE has
funded over 40 epidemiological studies of radiation effects on workers at
sites in the U.S. nuclear weapons complex. According to DOE, the results
have shown elevated cancer levels from chronic exposure at some sites, among
the most highly exposed workers, although the results have been inconsistent
and, looking complex-wide, DOE has not found a clear pattern of excess risk
for any specific cancer type. In general, epidemiological studies of workers
and other groups have the limitation of attempting to follow research
populations that may be too small to give statistically powerful or
conclusive results about very low radiation doses, and sometimes the studies
do not follow populations over a long enough period of time, according to
researchers and agency officials.

Epidemiological studies have been involved in the controversy over low-level
radiation effects because many such studies have attempted to statistically
correlate natural background radiation levels (principally radon) in the
United States and around the world with local cancer rates. A premise in the
controversy is that in places with higher background radiation levels, the
studies should show elevated cancer rates if there is a linear relationship
between low-level exposures and health risks. Some of the studies have
focused specifically on areas with the highest natural background radiation
levels. In some places, these levels are four or five times as high as
average U.S. levels. Two types of studies have been conducted. One type
relates regional rates of disease, often obtained from published vital
statistics, to measures of regional radiation levels. These are called
ecologic studies. The other type obtains disease data by direct interviews
of individuals or their representatives and by direct measurement of
radiation exposures. These are called analytic studies. Analytic studies can
be prospective, following populations and current exposures for future
outcomes, or retrospective, comparing current outcomes to past exposures.

For this review, we hired a consultant, Dr. Thomas Gesell, Professor of
Health Physics, Idaho State University, a recognized expert in the field of
environmental radiation, to identify and summarize worldwide ecologic and
analytic studies of natural background radiation or radon. Through his work,
we found that many ecologic and analytic studies have been done in the
United States, Europe, Asia, and South America. Some focused mainly on radon
effects, and others focused more broadly on overall natural background
radiation effects. The results of such studies differ and are inconclusive
overall. Most showed no evidence of elevated cancer risk, but a minority did
show slightly elevated cancer risks. Taken together, the studies may suggest
that low-level radiation effects are either very small, or nonexistent.
Scientists have disputed the importance of such studies in determining
low-level radiation effects. A factor in the dispute is the methodological
difficulty of performing meaningful epidemiological studies of cancer rates
in populations exposed to chronic low-level radiation doses. In many cases
the population being studied is limited in size, and the cancer effects
being pursued are small, making them difficult to detect among all cancers
in the population.

In most places on earth, natural background radiation (excluding radon)
varies within about a factor of four, and cosmic radiation varies by about a
factor of two in the elevations where most of the world's population lives.
Average world natural background radiation levels are about 240 millirem a
year (including radon), and about 300 millirem a year in the United States.
Particularly high concentrations of natural background radiation have been
reported in Brazil, India, China, and Iran. For example, there are areas of
elevated levels in Brazil, one along the coast (Guarapari) and another in
the interior (state of Minas Gerias); in the Kerala area, India; in
Yangjiang County, southwest Guangdong Province, China; and in Ramsar,
Caspian coast, Iran. In some of these areas, mean annual doses can be more
than double average U.S. levels, and hot spots into several thousand
millirem a year have been reported. In the United States, natural background
levels are over three times higher in the Rocky Mountains than along the
Gulf Coast.

With the help of our expert consultant, we examined 82 ecologic and analytic
studies of natural background radiation or radon, in the United States and
around the world. Of these studies, 45 were directly radon related. The
studies examined a variety of different types of cancer, and some examined
cancer effects on children, while others examined genetic effects. Results
of the studies varied, and we did not independently assess their quality.
Some reported statistically significant results--elevated cancer rates, no
elevation in rates, or a negative correlation--and others reported
inconclusive results. (Some lacked basic information for assessing their
quality.) Of 67 radon-related cancer studies, 22 reported results indicating
a statistically significant correlation between natural background radiation
or radon and cancer rates, while 45 found no such correlation (including 8
that found a negative correlation), and 4 were inconclusive. Others reported
statistically significant chromosomal aberrations in subjects, but not
cancer correlations.

In 1999, the National Academy of Sciences issued a report on the health
effects of radon exposure, called BEIR VI. Of 39 studies examined in the
report, including 19 ecologic studies, 17 studies reported a positive
correlation between radon and lung cancer, and 15 reported no correlation.
Three reported a negative correlation, including a 1995 ecologic study by
Bernard Cohen, University of Pittsburgh, of radiation-cancer correlations in
1,601 U.S. counties. The study found a strong tendency for lung cancer rates
to decrease with increasing radon exposures, in sharp contrast to the
increase expected from the linear model. The study and follow-up analysis
took steps to test for over 500 potential confounding factors, including
smoking and other lifestyle factors. The author found no potential
explanation for the discrepancy other than failure of the linear model. Some
critics of the study consider it to be carefully and thoroughly done but
believe that it lacks adequate data to control for smoking or other factors.
Many epidemiologists consider all ecologic studies to be methodologically
inferior to analytic case control studies because the former compile average
statistics, not individual statistics. Some analytic case control studies
have contradicted the Cohen study, including a 1997 meta-analysis (study of
studies) involving eight individual analytic studies of radon and lung
cancer. The meta-analysis found a small positive correlation between indoor
radon levels and lung cancer. However, the meta-analysis also found
considerable variance in the eight studies' results, and specific lifestyle
or socioeconomic confounding factors that would directly negate the Cohen
study's results have not been isolated, although various theories have been
put forward.

Costs of Different Radiation Standards

We examined numerous EPA, NRC, and DOE analyses of different nuclear cleanup
options and costs. The analyses generally confirmed that, as might be
expected, compliance costs varied to achieve different radiation protection
levels and accelerated to achieve more restrictive levels. The analyses were
done between 1991 and 1999, and they estimated costs under varying scenarios
to achieve different dose, risk, and contamination levels.37 In particular,
we examined several site-specific DOE nuclear cleanup studies; an NRC
environmental impact study of four types of generic NRC-regulated
facilities, performed to support NRC's decommissioning standard; and a draft
EPA regulatory impact analysis of the potential nationwide costs (at 16
generic types of sites) of EPA's proposed comprehensive cleanup standards.
Selected analyses that showed cost variations in relation to achieving
different millirem levels are discussed below.

Our examination of DOE's and NRC's analyses showed potential site-specific
cost differences of millions of dollars, in some cases, between cleaning up
radioactively contaminated soil to 15 millirem a year and cleaning it up to
25 millirem a year, under various scenarios. (These analyses do not
represent estimates of overall site cleanup costs, which may include factors
in addition to soil cleanup, including remediation activities such as the
decontamination and removal of contaminated equipment and structures and
liquid waste treatment.) EPA's analysis showed potential nationwide
incremental costs in the low billion dollars to achieve more restrictive
cleanup levels.38 Major analyses that showed cost variations in the range of
100 millirem a year and below are shown in table 2.

 Dollars in millions
 Agency/site/analysis
 date                  Estimated soil cleanup cost
                                                                                Less than
                       100-millirem-a-year 25-millirem-a-year 15-millirem-a-year10-millirem-a-year
                       level               level              level             level
 DOE
 Hanford process
 waste areas, 1994a    b                   14                 19                59c
 Nevada Test Site and
 test ranges, 1995d    35                  131                240               1,003e
 Weldon Spring, 1991f  b                   0.58               1.4g              2.0h
 Brookhaven waste
 facility, 1998i       15.9                24.4               28.2              64.5j
 NRCk
 Generic nuclear
 power plant, 1997     0.17                0.31               0.41              1.44c
 Generic fuel
 fabrication           2.89                6.35               7.83              13.56 c
 facility, 1997
 Generic metal
 extraction facility,  5.30                6.21               7.33              13.86 c
 1997
 Generic sealed
 source facility,      0.08                0.26               0.35              0.64 c
 1997
 EPA
 Generic nationwide
 sites, 1996l          m                   1,000              1,500             3,200c

aThe analysis was part of a feasibility study for a Superfund cleanup of
ponds, trenches, and burial grounds in Area 300, where uranium and other
wastes from processing nuclear fuel for Hanford's production reactors were
disposed of over decades. Under an industrial scenario, the analysis
considered five cleanup options. The totals shown are for the selected
cleanup and disposal option and are present values.

bCost not calculated for this level.

c3 millirem a year.

dAt the site and test ranges (Nellis and Tonapah), plutonium contaminates
surface soils over an estimated 90,000 hectares (at levels above 10
picocuries per gram). The contamination resulted from atmospheric nuclear
tests, as well as safety tests that subjected nuclear devices to
conventional explosives. The analysis was conducted separately from the
Superfund process, as a cost-risk-benefit case study to evaluate a range of
cleanup alternatives, using an active institutional controls scenario to
limit costs. An integration model was used to estimate individual risks,
population risks, costs (cleanup costs, costs of worker fatalities), and
benefits (including reduced future on-site cancer fatalities). The totals
assume on-site disposal for the Tonapah range waste.

e5 millirem a year.

fThe analysis supported the Superfund process for the site, in Missouri,
where uranium and thorium ore were processed from 1957 until 1966,
contaminating buildings and soil. Under an on-site farmer scenario,
alternative uranium soil cleanup levels were calculated. A
25-millirem-a-year cleanup target was selected, with an as low as reasonably
achievable (ALARA) goal equivalent to 6.7 millirem a year.

g12 millirem a year.

h6.7 millirem a year.

iThe facility historically processed, treated, and stored radioactive and
hazardous waste. The analysis was conducted apart from the Superfund
process, as an ALARA case study. Using an open space scenario, the analysis
undertook risk optimization to determine the cost savings from the remedial
action, the life-cycle costs for the remedial action, the value of the net
dose avoided, and the value of other avoided risks and damages. In the net
benefit-cost analysis, net costs resulted at all millirem-per-year levels.
The totals shown represent present values.

j1 millirem a year.

kThe four generic types of NRC-liensed facilities were included in a generic
environmental impact statement supporting the issuance of the agency's
decommissioning standard. The generic environmental impact statement
included analyses of human health impacts, costs, and the cost-benefit of
the regulatory action, under an unrestricted-use scenario. The totals shown
reflect soil removal and survey costs, based on "real world" soil
characterization estimates.

lThe analysis was part of a regulatory impact analysis supporting EPA's
prospective nuclear site cleanup standard. The analysis addressed
incremental soil cleanup cost and health impact differences nationwide,
between a base case (100 millirem a year, absent the proposed EPA standard)
and the regulatory case (assuming promulgation of the EPA standard) for 16
types of reference sites, based on actual DOE, DOD, and NRC-licensed sites.
Results were calculated for three scenarios--upper- and lower-bound
scenarios and an intermediate scenario representing EPA's best estimate. The
totals shown reflect the intermediate case, representing incremental soil
cleanup costs below a 100-millirem-a-year baseline. The totals represent
present values.

mThe 100-millirem-a-year baseline.

As shown in table 2, for individual DOE and NRC-licensed sites, the
estimated differences in compliance costs between the 15-millirem-a-year and
the 25-millirem-a-year levels--reflecting EPA's and NRC's all-pathway
standards, respectively--varied widely, from thousands of dollars to over
$100 million. Similarly, the table shows widely varying cost differences
between the 100-millirem-a-year level and lower levels. EPA's analysis
estimated incremental compliance costs for multiple sites nationwide. As
shown in the table, the analysis found cost differences of over a billion
dollars between the 100-millirem-a-year level and lower levels.

The varied costs in table 2 reflect not only different protection levels but
also different land-use scenarios and site characteristics, among other
factors. For example, the 1995 analysis for the Hanford process waste areas
used an industrial scenario, at a location with a variety of soil and
structure components; the 1995 analysis for the Nevada test ranges used an
active institutional controls scenario, at a large, soil dominated location;
and NRC's 1997 analysis used unrestricted access scenarios, at generic
industrial sites with substantial on-site structures. According to DOE
analysts, scenarios involving on-site unrestricted, residential, or
industrial

use can result in much lower cleanup levels and higher cleanup costs than
scenarios involving restricted or open-space use.39

The analyses also generally showed that cleanup costs accelerated to achieve
the most restrictive protection levels. These accelerating costs can be
shown graphically in the form of a cost-benefit curve, as depicted in figure
3, for the Brookhaven hazardous waste facility:

As shown in figure 3, in the Brookhaven analysis, costs were a function of
cleanup levels, increasing gradually, from right to left, from 100 millirem
a year to 15 millirem a year, and accelerating from 15 millirem a year to 1
millirem a year. Conversely, rates of dose reduction (and potential
associated increased health benefits) decelerated at costs above $30
million. This relationship generally held for many DOE analyses we examined,
including some that expressed cleanup levels in terms of risk and others
that expressed cleanup levels in terms of concentration levels achieved.40
According to DOE analysts, under the principle of reducing doses to levels
as low as reasonably achievable (ALARA), there is a point or "elbow" on the
cost-curve where an optimal balance between costs and cleanup level is
chosen.

Agencies generally did not have overall estimates of the cost differences
between achieving NRC's all-pathway 25-millirem-a-year limit for nuclear
cleanups and achieving EPA's proposed extra groundwater protection
requirements, based on the agency's more restrictive drinking water
standards. However, available DOE and NRC analyses showed potential
multibillion dollar cost differences per site to achieve EPA's standards.
Both NRC and DOE expressed concern that the extra groundwater protection
favored by EPA for nuclear sites could impose multimillion dollar additional
costs annually at sites with groundwater contamination. NRC, in part, based
its concern on an analysis done to support its decommissioning standard.
This analysis showed that at a generic nuclear site with strontium 90
groundwater contamination, from a baseline of 25 millirem a year,
incremental costs using "pump and treat" methods could be as high as $7
million to achieve a 3-millirem-a-year level (below the original 4
millirem-a-year drinking water standard, but above the level for strontium
90 using up-to-date dose estimation methods--0.07 millirem a year) and an
additional $32 million to achieve natural background levels. ("Pump and
treat" means that the water is pumped out of the ground, treated by various
means, and discharged back into the ground.)

Site-specific DOE analyses also indicated incremental costs of many millions
of dollars per site to meet drinking water standards through long-term pump
and treat techniques. For example, a 1998 DOE analysis, issued in concert
with EPA and the Idaho Department of Health and Welfare, showed options for
cleaning up the Snake River Plain Aquifer underneath the spent fuel
processing facility at the Idaho National Engineering and Environmental
Laboratory. According to the analysis, the costs of using a pump and treat
approach, if necessary to achieve the drinking water standard for a key
radionuclide, Iodine 129, by 2095, could be anywhere from about $40 million
to about $788 million depending on the aggressiveness of the approach
taken.41

Also, a 1998 Brookhaven National Laboratory analysis of pump and treat
options to clean up tritium and strontium 90 in the on-site groundwater
showed that natural attenuation of the strontium 90 for 60 years or more
might meet the drinking water standard for strontium 90 and cost less than a
million dollars for monitoring activities. Alternatively, pump and treat
methods might achieve the drinking water standards in 30 or more years, at
costs of from $5.8 million to $6 million.

Comments From the Environmental Protection Agency

The following are GAO's comments on the letter dated June 13, 2000, from the
Director, Office of Radiation and Indoor Air, Environmental Protection
Agency.

1. EPA said the draft report raised the issue of whether the U.S. government
should continue to use the linear model, and EPA said it is following the
consensus of scientific organizations in doing so. EPA also said the lack of
scientific evidence has not led to the agency's regulatory disagreement with
NRC. Rather, according to EPA, the two agencies have made different
risk-management decisions, based on differing statutory mandates. One of our
report's objectives was to describe the scientific basis for U.S. radiation
standards, which we found to be inconclusive. The report raises no
expectation that the linear model will or should be soon superseded, pending
the existence of better evidence of low-level radiation effects. We agree
that risk management decisions made by EPA and NRC, based on differing
statutory mandates and protective strategies, largely account for the
disagreement between the two agencies. Our report says this.

2. EPA said that in citing low-level radiation as a weak carcinogen, GAO
ignored evidence that ionizing radiation has a much higher probability of
resulting in DNA misrepair than other, more commonplace cell-altering
events. We are aware of such evidence, and in agreement with EPA's
qualifying point, we modified the final report accordingly.

3. EPA said that its groundwater standard is not alone responsible for
driving the cost of the Yucca Mountain project, and EPA's standards have had
little time to influence the overall cost of the program. Our report
mentions several factors, significantly but not exclusively including EPA's
proposed groundwater standard, that have influenced the project's costs to
date and could do so in the future.

4. EPA said it is premature to conclude that a memorandum of understanding
clarifying EPA's and NRC's regulatory roles is unlikely. We believe that,
given the lack of progress by the two agencies since as long ago as 1992, it
is questionable whether they will finalize such a memorandum. Even if they
do, we question whether such a finalized memorandum will fundamentally
resolve the problem between them.

5. EPA said the draft report should have focused more on the relatively
small differences in cleanup costs between the 25-millirem-a-year and
15-millirem-a-year levels, which EPA and NRC have often argued about. The
scope of our review did not exclude consideration of levels both higher and
lower than 25 millirem a year and 15 millirem a year. The report shows that
cost differences to achieve these levels vary, depending on the site
involved, and may be small in some cases. However, in relation to the
hundreds of potential cleanup sites that exist nationwide, overall cost
differences between the 25-millirem-a-year and 15-millirem-a-year levels
could be substantial.

The following are GAO's comments on the letter dated June 12, 2000, from the
Director, Office of Emergency and Remedial Response, Environmental
Protection Agency.

1. EPA said the draft report appears to evaluate radiological site
contamination and cleanup issues without sufficient acknowledgement that
chemical contamination also exists at almost all of these sites. Although
the subject of our report is radiation protection, the report states that
EPA's policy is to coregulate chemicals and radionuclides, within the same
risk range, rather than to treat them differently. EPA also said its risk
management approach for cleaning up sites is generally consistent with that
of other agencies, including DOE (but not NRC). Although our report focused
mainly on regulatory differences between the two principal standard-setting
agencies, EPA and NRC, the report points out that NRC and DOE (in its worker
protection standards and proposed public protection standards) generally
favor a "top down" approach of setting a relatively less restrictive dose
limit, and then reducing doses well below the limit in site-specific
situations. On the other hand, EPA's Superfund approach has been "bottom
up," setting a relatively restrictive risk goal but allowing less
restrictive limits in site-specific situations.

2. EPA said the draft report does not adequately discuss recent EPA-NRC
initiatives to avoid potential dual regulation at cleanup sites and
inaccurately portrays the two agencies as making little progress in
negotiating a memorandum of understanding. Our draft report refers to these
recent initiatives, but we modified the final report to make more specific
mention of them. Overall, in relation to the EPA-NRC disagreement and these
initiatives, our report essentially takes the longer view, noting that since
as long ago as 1992, the two agencies have been unsuccessful in addressing
their potentially conflicting roles in site cleanups. On this basis, it is
unclear that the latest initiatives will succeed without congressional
intervention. EPA also said the draft report fails to mention the
administration's opposition to proposed legislative changes, such as
amending Superfund legislation, in lieu of the memorandum of understanding
approach. While we are aware of past legislative proposals, our report does
not delineate what form congressional clarification of the two agencies'
regulatory responsibilities might take.

3. EPA doubted that nuclear site cleanup costs would begin to dramatically
rise at the same cleanup level for different sites. Our report makes no such
assertion. In general, the agency analyses we examined showed faster rising
costs at lower cleanup levels.

Comments From the Nuclear Regulatory Commission

Comments From the Department of Energy

The following are GAO's comments on the Department of Energy's letter dated
June 13, 2000.

1. DOE raised the question of references to information sources in our draft
report. Our report, for the most part, does not cite technical sources, in
keeping with GAO's policy and the role of our reports as other than
technical treatises, as well as in the interest of report brevity.

2.DOE said there may be different interpretations of the "no backsliding"
provision of the 1996 Safe Drinking Water Act Amendments. We believe that
groundwater protection policy and legal matters, including the specific "no
backsliding" provision, will be of interest to the congressional committees
of jurisdiction in any efforts they may wish to undertake to reconcile EPA's
and NRC's regulatory approaches.

3. DOE said the draft report did not address the integrated health risk or
detriment associated with or averted by the standards and associated cleanup
levels, and it pointed out that risk reductions associated with cleanup
efforts may provide little if any health benefit. Our report recognizes that
agencies routinely calculate hypothetical cancer deaths averted in achieving
cleanup levels. However, we regard such calculations with caution because
they are based on the linear model, and therefore we did not highlight such
calculations in our report.

(141363)

Table 1: Estimated Costs to Achieve Different Soil Cleanup
Levels at Selected DOE Sites and Generic
NRC-Licensed Sites 27

Table 2: Potential Costs to Achieve Different Soil Cleanup
Levels--DOE's, NRC's, and EPA's Analyses 41

Figure 1: The Linear, No-Threshold Model of Low-Level
Radiation Effects 11

Figure 2: Four Models of Low-Level Radiation Effects 35

Figure 3: Cleanup Costs as a Function of Cleanup
Levels--Hazardous Waste Facility, Brookhaven
National Laboratory, 1998 43
  

1. Nuclear Health and Safety: Consensus on Acceptable Radiation Risk to the
Public Is Lacking (GAO/RCED-94-190 , Sept. 19, 1994).

2. These means of exposure, called "pathways" by specialists, include
exposure through soil, water, and air.

3. A millirem is a commonly used unit of measurement of the biological
effect of radiation. The radiation from a routine chest X-ray is equivalent
to about 6 millirem.

4. Above about 30,000 total millirem, radiation exposure is a well-known
cause of cancer. Instantaneous (or short-duration) exposures of about
200,000 total millirem can cause blood cell changes, infections, and
temporary sterility. Short-duration exposures above about 400,000 total
millirem can cause death within days or a few weeks and are associated with
catastrophic nuclear accidents or atomic bomb blasts.

5. U.S. worker protection standards limit annual exposures to 5,000
millirem. See app. II.

6. About 187,000 spontaneous cell-altering events occur daily in each human
cell. Low-level radiation exposures increase the number of such events by a
small fraction--about 1 percent. There is evidence that the type of damage
done by such radiation has a higher probability of resulting in DNA
misrepair than the type of damage done by other normally occurring
cell-altering events.

7. Superfund regulations call for, among other steps, the development of
cleanup alternatives through a remedial investigation, the finalization of
applicable cleanup requirements, and a formal record of decision on an
agreed cleanup level, after which cleanup is conducted. Decommissioning
involves the removal of all radioactive components and materials from the
facility and the cleanup of radionuclides to NRC's standards (10 C.F.R. 20).

8. The coordination of federal radiation protection issues is the
responsibility of the federal Interagency Steering Committee on Radiation
Standards, and coordination among states is the responsibility of the
Conference of Radiation Control Program Directors. In addition,
authoritative national and international technical organizations make
recommendations on radiation protection issues, including the
congressionally chartered U.S. National Council on Radiation Protection and
Measurements, the International Commission on Radiological Protection, and
the United Nations Scientific Committee on the Effects of Atomic Radiation.

9. EPA's protection approach draws, in part, on experience with regulating
thousands of different chemicals, many of which pose risks that are
generally thought to be even less well understood than radiation risks.
These chemicals may or may not exist naturally in the environment.

10. NRC's approach (with which DOE generally concurs) draws on experience in
estimating radiation-specific risks, within an internationally recommended
radiation dose limitation and risk assessment framework. The framework takes
into account that radiation exists naturally in the worldwide environment.

11. NRC's 25-millirem-a-year dose limit is equivalent to 1 chance in about
1,000 of a fatal cancer over a 70-year lifetime, using a commonly accepted
dose-risk conversion factor and assuming the linear model of radiation
effects holds.

12. For example, a 1990 study by a National Academy of Sciences committee,
called BEIR V, estimated that, at the 90-percent statistical confidence
interval, out of 100,000 adults exposed to 100 millirem a year of radiation
over a lifetime, anywhere from 410 to 980 men and 500 to 930 women might die
of cancer caused by the exposure. This confidence interval assumes the
validity of the linear model and reflects the uncertainty of inputs to the
model.

13. There is evidence of mental retardation linked to fetal exposure to
low-level radiation. Age at time of exposure appears to be an important
determinant of cancer risk from radiation.

14. The British National Radiation Protection Board similarly maintains, in
consideration of relevant cellular and molecular data, that the weight of
evidence falls decisively in favor of the thesis that for a majority of
common human tumors, low-dose and low-dose-rate cancer risk rises as a
simple function of dose with no threshold.

15. According to one expert, extrapolating effects from high exposures to
low exposures equivalent to natural background radiation levels is more
guesswork than science.

16. Such research focuses on cells' nuclei, where DNA is located.

17. BEIR VI was a 1999 Academy assessment of risks from radon.

18. EPA's approach includes various levels of acceptable risk, from 1 chance
in about 2,000 to less than 1 chance in about a million of a fatal cancer
over a 70-year lifetime. NRC's all-pathway approach involves a level of
acceptable risk of 1 chance in about 1,000 of a fatal cancer over a
lifetime. These calculated risks are based on a commonly used dose-risk
conversion factor, assuming the linear model holds.

19. The agencies also disagreed on low-level waste disposal standards in the
mid-1990s. Current low-level waste standards consist of NRC's
25-millirem-a-year all-pathway limits that date from 1983. In 1994, EPA
considered issuing its own standards, reflecting 15-millirem-a-year
all-pathway protection, plus extra groundwater protection to drinking water
standards. At the time, DOE estimated over $300 million in added annual
costs if its disposal sites and commercial disposal sites were required to
comply with the approach EPA was considering.

20. In addition, in 1994, EPA issued transuranic waste disposal standards
for the Waste Isolation Pilot Project in southeastern New Mexico (40 C.F.R.
191) that include 15-millirem-a-year all-pathway limits, plus extra
groundwater protection to drinking water standards. NRC expressed concerns
about the groundwater protection standards but concurred with them because
on-site groundwater was not an issue in EPA's project certification
process--the aquifer was brine. Transuranic waste is tools, rags, laboratory
equipment, and other items contaminated with radioactive elements, mostly
plutonium.

21. Under 40 C.F.R. 141, annual concentrations of beta particle and photon
activity sources are limited to no more than a total body or internal organ
dose equivalent of 4 millirem a year. See app. II.

22. On the other hand, the Academy found the magnitude of EPA's proposed
15-millirem-a-year all-pathway limit to be consistent with the Academy's own
recommendations.

23. For example, the limit for Iodine 129, considered a benchmark among the
various limits for Yucca Mountain, is 1 picocurie per liter, or about 0.2
millirem a year; the limit for Nickel 63 is 50 picocuries per liter, or
about 0.02 millirem a year, and the limit for Tritium is 20,000 picocuries
per liter, or about 0.9 millirem a year. In addition, the limits reflect
acceptable lifetime risks ranging anywhere from less than 1 chance in a
million to more than 1 chance in 2,000 of a person dying from the exposure,
using a commonly accepted dose-risk conversion factor and assuming the
linear model holds. EPA points out that most limits fall within its
acceptable risk range of 1 chance in about 10,000 to 1 chance in about a
million of a person getting cancer from the exposure.

24. Another technically related groundwater issue is EPA's prospective
choice of a groundwater scenario for Yucca Mountain, including the point of
enforcement (at or how near to the repository boundary) and appropriate
estimated groundwater flow volume. DOE and NRC officials said a very
conservative scenario could severely complicate DOE's efforts to do
detailed, refined groundwater analysis for the site.

25. According to EPA, the new limits are to be based on acceptable risks
instead of the current dose basis of 4 millirem a year.

26. In addition, DOE has issued public protection orders for its nuclear
installations that generally conform to NRC's approach, including
all-pathway protection without extra groundwater protection, as well as dose
reductions to levels as low as reasonably achievable. DOE has proposed to
convert its order into a regulatory standard, but EPA has opposed the draft
standard as inconsistent with Superfund requirements.

27. In 1994, NRC considered standards comparable to EPA's--15 millirem a
year, with separate groundwater protection to drinking water standards--but
changed to an all-pathway, 25-millirem-a-year approach after further
analysis and public comments on the proposed rule.

28. Also, over the years, differences between EPA and DOE concerning
standards and acceptable risks for cleanups at DOE sites have contributed to
regulatory delays and higher regulatory and cleanup costs while raising
public questions about what cleanup levels are appropriate. See Nuclear
Cleanup: Completion of Standards and Effectiveness of Land-Use Planning Are
Uncertain (GAO/RCED-94-144 , Aug. 26, 1994).

29. The licensee has stated that it can meet more stringent standards of 10
millirem a year, plus extra groundwater protection to the equivalent of 4
millirem a year, as imposed by the state of Maine.

30. Annual costs and benefits of environmental regulations have been
estimated to total many billions of dollars annually. For example, see R.
Hahn, and J. Hird, "The Costs and Benefits of Regulations," Yale Journal on
Regulation, vol. 8 (1991), pp. 233-78.

31. DOE officials said that since the 1990s, they have been designing the
repository to more than meet prospective radiation standards that EPA and
NRC might issue.

32. At another nuclear waste disposal facility that is already in operation,
the Waste Isolation Pilot Project, where EPA's transuranic waste disposal
standards are operative, DOE projects funding to 2070 at $7.7 billion. In
addition, state compacts and unaffiliated states have to date incurred
almost $600 million in costs for planning and developing potential low-level
waste disposal sites, although no sites have been built. See Low Level
Radioactive Wastes: States Are Not Developing Disposal Facilities
(GAO/RCED-99-238 , Sept. 17, 1999).

33. According to DOE and EPA, NRC's proposed "reasonable assurance"
performance objective for the repository may be more stringent and costly to
implement than the "reasonable expectation" compliance objective in EPA's
proposed standards. NRC disagreed that this would necessarily be the case.

34. See Nuclear Waste: DOE's Accelerated Cleanup Strategy Has Benefits but
Faces Uncertainties (GAO/RCED-99-129 , Apr. 30, 1999). According to DOE,
about 85-90 percent of its environmental management budget is directed
toward ensuring compliance with the large number of legally enforceable
cleanup and compliance agreements in place at major sites around the
country. Such compliance involves not only EPA requirements, such as
Superfund, but also any applicable state requirements, as well as DOE's own
radiation protection orders.

35. An exception is a preliminary regulatory analysis done by EPA in 1996 to
support its proposed cleanup standards. The analysis showed incremental cost
differences in the low billions of dollars to meet various cleanup levels
below 100 millirem a year. See app V.

36. These analyses do not consider overall site cleanup costs, which may
include many factors, such as the costs of decontaminating and removing
structures and treating liquid waste. The analyses often estimated
hypothetical cancer deaths averted from meeting various protection levels.

37. Often, in addition to costs, the analyses estimated doses and cancer
deaths averted from meeting various protection levels, using the linear
model.

38. EPA quantified soil cleanup costs in the draft analysis. According to
DOE's comments on the analysis, EPA may have greatly underestimated the
potential costs of implementing EPA's drinking water standards for
groundwater at DOE sites.

39. For example, the choice of an open space scenario for the Rocky Flats
facility in Colorado was a factor in the decision to apply a less
restrictive plutonium cleanup level there than was required at the Nevada
Test Site. See DOE: Accelerated Cleanup of Rocky Flats--Status and Obstacles
(GAO/RCED-99-100 , Apr. 30, 1999). The Rocky Flats plutonium cleanup level
is being reconsidered in response to stakeholders' concerns, and a
preliminary DOE cost estimate for a many times more restrictive cleanup
level showed potential cleanup cost increases for Rocky Flats in the tens of
millions of dollars.

40. Other analyses included a draft 1999 analysis at two Nevada Test Site
locations in the Tonapah test range; a 1999 analysis at the Energy
Technology and Environmental Center, Santa Susana, California; a 1995
analysis at the Elza Gate, Tennessee, former waste storage site; and a 1994
analysis at the Ventron, Massachusetts, former uranium compounds processing
facility.

41. According to a DOE-Idaho official, a December 1999 record of decision
based on the analysis chose a less expensive aquifer dilution and monitoring
approach.
*** End of document. ***