LOGAN, Circuit Judge.
These consolidated cases involve challenges to the Environmental Protection Agency’s (EPA) standards for the cleanup and disposal of uranium mill tailings originating from designated inactive mill sites. The EPA established these standards pursuant to its authority under the Uranium Mill Tailings Radiation Control Act of 1978 (UMTRCA), as amended, 42 U.S.C. §§ 2022 and 7901-7942. The UMTRCA required the EPA to promulgate standards that could be applied generally to protect the environment and the public health and safety from radioactive and nonradioactive hazards posed by uranium mill tailings at both active and inactive processing sites. Under the statutory scheme the federal government and the affected state share the costs of the remedial action taken to control mill tailings, see 42 U.S.C. § 7917, with the possibility of later reimbursement from private parties, id. § 7925. The EPA standards that we review here pertain only to inactive mill sites. In a companion case released this day we review regulations relating to active mill sites. See American Mining Congress v. Thomas, 772 F.2d 640 (10th Cir.1985) (Active Sites Case).
The following parties filed petitions for review of these standards: the American Mining Congress, a trade association; joint petitioners United Nuclear Corporation, Kerr-McGee Corporation, Kerr-McGee Nuclear Corporation, and Homestake Mining Company; joint petitioners Sierra Club, Environmental Defense Fund, Natural Resources Defense Council, Southwest Research and Information Center and Jean Slattery; and the State of Colorado as in-tervenor. In addition, the State of Wyoming filed an amicus brief.
I
The final product of the milling process for uranium ore is uranium-rich “yellow-cake,” U3O8. The milling process also produces a residue of either slime or coarse sand. This residue, which comprises the uranium mill tailings piles, contains radioactive material, the most significant of which is radium. Radium decays to produce radon. Radon is an inert gas, some of which escapes from the tailings particles into the atmosphere. Airborne radon degrades into a series of short half-life decay products that are hazardous if inhaled. If the radon gas does not escape the mill tailings piles, its decay products remain in the piles and produce gamma radiation, which may be harmful to people and animals living near the mill tailings piles. Uranium mill tailings also contain potentially dangerous nonradioactive materials such as arsenic and selenium. These toxic and radioactive materials may be ingested with food or water. 48 Fed.Reg. 590, 592 (1983). See generally I Environmental Protection Agency, Final Environmental Impact Statement for Remedial Action Standards for Inactive Uranium Processing Sites 3-68 (1982) [hereinafter FEIS-IN],
To deal with the perceived dangers presented by uranium mill tailings, Congress enacted the UMTRCA. When it passed this legislation in 1978, it stated, in a section titled “Congressional findings and purposes”:
“uranium mill tailings located at active and inactive mill operations may pose a potential and significant radiation health hazard to the public, and that the protection of the public health, safety, and welfare ... require[s] that every reasonable effort be made to provide for the stabilization, disposal, and control in a safe and environmentally sound manner of such tailings in order to prevent or minimize radon diffusion into the environment and to prevent or minimize other environmental hazards from such tailings.”
42 U.S.C. § 7901(a).
In the UMTRCA, Congress gave the Department of Energy (DOE) and the Nuclear Regulatory Commission (NRC) responsibility for implementing a remedial program to clean up and dispose of the mill tailings. See id. §§ 7911-7924. The EPA is'responsible for promulgating the general standards that the implementing agencies must meet. See id. §§ 2022(a), 7918(a).
[622]*622The EPA issued proposed general standards for the remedial program in two parts: cleanup standards and disposal standards. The EPA intended the cleanup standards to reduce the detrimental health consequences of tailings that have been dispersed from the tailings piles or used in construction. 45 Fed.Reg. 27,370, 27,370 (1980). The EPA intended that the disposal standards place the tailings piles “in a condition which will be safe for a long time.” Id.
On April 22, 1980, the EPA Administrator published for comment “Proposed Cleanup Standards for Inactive Uranium Processing Sites.” 45 Fed.Reg. 27,370 (1980). These proposed standards were “for the cleanup of open lands and buildings contaminated with residual radioactive materials (mainly tailings) from inactive uranium processing sites.” Id. The EPA made them immediately effective as interim standards pending comment, review, and promulgation of the final standards. The EPA also issued a draft environmental impact statement to support the proposed standards.
The proposed cleanup standards established allowable levels of radium concentration in soil contaminated by dispersed tail-ings. The standards also set permissible levels of radon decay product concentration and gamma radiation in occupied or occupiable buildings affected by the tailings.1
In addition, these proposed standards contained exceptions to strict compliance if certain criteria were met. Id. at 27,375. At qualifying sites, the implementing agency was to perform remedial action that would come as close as possible to meeting the standard to which the exception applied. Id.
The EPA issued the second set of proposed standards, the disposal standards, on January 9, 1981. 46 Fed.Reg. 2556 (1981). The disposal standards placed limits on the radon release to the atmosphere from the tailings piles and also placed limits on water contamination from the piles. Id. These standards required that the tailings be disposed of in a manner “that provides a reasonable expectation that these limits will be satisfied for at least one thousand years.”2 Id.
In the proposed disposal standards the EPA left little doubt that it foresaw covering the tailings piles as the most viable means to achieve the proposed radon emission standards. The EPA stated,
“In the draft EIS we analyze the health and environmental protection benefits and the costs of several levels of controlling tailings, assuming a variety of potential control methods. We find that [623]*623radon emission levels of an ‘average’ pile can be reduced to approximately the levels characteristic of ordinary land by applying a soil cover at costs in a range of about 1 to 14 million (1979) dollars.”
Id. at 2559.
One of the standards established limits for concentrations of toxic substances in underground sources of drinking water.3 Another water standard required that substances released from the disposal site “after disposal will not cause the concentration of any harmful dissolved substance in any surface waters to increase above the level that would otherwise prevail.” Id. at 2562.
The EPA received extensive comments on both the proposed cleanup and disposal standards. See II FEIS-IN (summary of comments and responses). The comments were wide ranging — the industry petitioners argued that there was insufficient evidence of risk of harm to warrant the standards, while the environmental groups argued that the proposed standards did not provide adequate protection against the risks posed by the tailings. Both sides buttressed their arguments with technical studies and expert analysis.
Congress in 1982 discussed amendments to the UMTRCA to extend the deadline for the promulgation of final standards and enacted an amendment on January 4, 1983. See Act of Jan. 4, 1983, Pub.L. No. 87-415, sec. 18, 96 Stat. 2067, 2077 (1983) (codified at 42 U.S.C. § 2022(a)) (generally referred to as the “1982 amendment”). Anxious to institute standards for the mill tailings, Congress also provided that should the EPA miss the extended deadline, remedial action would commence using the proposed standards. Id.; House Conf.Rep. No. 884, 97th Cong., 2d Sess. 44-45, reprinted in 1982 U.S.Code Cong. & Ad.News 3592, 3603, 3614-15. By the time Congress passed the amendment its specified deadline for inactive sites had already passed. See 42 U.S.C. § 2022(a). Important for our purposes is that Congress also included in the 1982 amendment a sentence explaining the various factors that it required the EPA to consider in developing the standards: “In establishing such standards, the Administrator shall consider the risk to the public health, safety, and the environment, the environmental and economic costs of applying such standards, and such other factors as the Administrator determines to be appropriate.” Id.
On January 5, 1983, one day after Congress passed the amendment, the EPA promulgated the final “Standards for Remedial Actions at Inactive Uranium Processing Sites,” which covered both cleanup and disposal requirements. 48 Fed.Reg. 590 (1983) (filed Dec. 30, 1982) (codified at 40 C.F.R. §§ 192.00-43 (1984)). The final regulations differed, in some ways quite substantially, from the proposed standards. [624]*624In explaining the changes in general terms the EPA said:
“In response to comments on the proposed standards for disposal and for cleanup, we have evaluated a number of alternatives in terms of their costs and the reductions achievable in potential health effects. A number of changes have been made, including raising some of the numerical limits and eliminating some requirements. The purpose of most of these changes is to make implementation easier and less costly. The changes should not result in any substantial loss of health or environmental protection over that which would have been provided by the proposed standards.”
48 Fed.Reg. at 590.
After specifically explaining in a table the various alternatives it had considered for disposal of the mill tailings piles, the EPA divided the alternatives into categories of least cost, optimized cost-benefit, or nondegradation alternatives. Id. at 591. To justify the less stringent final disposal standards, the EPA said that the original standards it had proposed were close to nondegradation standards and that they would have been difficult to implement because they required cleanup and control limits close to background levels of radiation. Id. It also said that “the small incremental health benefits, when compared to the benefits for less stringent alternatives, do not appear to justify the large additional costs.” Id. In its final standards, the EPA purported to select an “optimized cost-benefit” rather than a “least cost” altentative. Id. The final standard for control or disposal of the tailings piles set the radon emission limit at 20 pCi/m2s rather than the proposed 2 pCi/m2s limit.4 Id. Furthermore, the EPA changed the flat 1000 year disposal longevity requirement, qualifying it with the language, “to the extent reasonably achievable and, in any case, for at least 200 years, ____” 40 C.F.R. § 192.02(a) (1984).
The final regulation set the cleanup standard for land, the maximum concentration level of radium 226, at 5 pCi/gram averaged over the first 15 centimeters of soil and at 15 pCi/gram for soil layers more than 15 centimeters below the surface.5 This standard was an increase from the 5 pCi/gram level set out in the proposed standard for subsurface soil. 46 Fed.Reg. at 2562. After explaining the differences between the proposed and final regulations for dispersed tailings contaminating the soil, the EPA stated that it believed the final standards “will result in essentially the same degree of cleanup, and will be simpler to implement.” 48 Fed.Reg. at 600.
The EPA also adjusted the final standards on the limits for radioactivity in buildings affected by the mill tailings. The EPA increased the maximum permissible level of radon decay product concentration from 0.015 WL to 0.03 WL, requiring, however, that “reasonable effort shall be made to achieve, an annual average (or equivalent) radon decay product concentration (including background) not to exceed 0.02 WL.” 40 C.F.R. § 192.12(b)(1) (1984). The [625]*625level of gamma radiation allowed by the final standards is no more than 20 microro-entgens per hour over background level. Id. § 192.12(b)(2).
Commenting upon the difficulty of assessing water contamination at the various sites and stating a belief that there was a low probability of additional contamination at most sites, the EPA retreated from its proposed water regulations, stating that it was refusing to enact general regulations on the subject. 48 Fed.Reg. at 599. The EPA, however, did recognize the potential for site-specific water contamination problems. In the cases where there was a potential for ground water contamination, the EPA declared that it had provided:
“in the implementation section of these standards, that judgments on the possible need for monitoring or remedial actions should be guided by relevant considerations described in EPA’s hazardous waste management system, and by relevant State and Federal Water Quality Criteria for existing and anticipated uses of the aquifer.”
Id. at 599-600. The EPA adopted the same approach to surface water contamination, leaving the regulatory burden to the DOE and the NRC on a site-specific basis, although it indicated that it thought surface water would be adequately protected in any case by its general disposal standards. Id.
The final standards, like the proposed standards, incorporated provisions allowing the implementing agency to grant exceptions if certain criteria were met. See 40 C.F.R. §§ 192.21-22 (1984).
The industry petitioners, the environmental groups, and the State of Colorado raise the following issues for our consideration: (1) whether the EPA must find that the mill tailings piles present “a significant risk” of harm before promulgating standards under the UMTRCA; (2) whether the EPA’s radon emission and radium-in-soil standards are invalid because they are on-site standards beyond the EPA’s authority to promulgate; (3) whether the EPA standards unlawfully impose management, design, and engineering requirements; (4) whether the EPA should engage in a cost-benefit analysis in establishing the standards for the final regulations, particularly the radon emission standard; (5) whether the EPA’s standards for the mill tailings piles are arbitrary and capricious because they are unsupported by the record and bear no rational relationship to the protection of the public safety and health and the costs required to implement them; and (6) whether by abandoning general water quality standards and standards regarding non-radiological toxic pollutants the EPA has acted contrary to the law.
The UMTRCA refers to the Administrative Procedure Act, 5 U.S.C. §§ 701-706, for standards of judicial review of rules promulgated under it. See 42 U.S.C. § 2022(c)(2). The appropriate standard of review for this type of informal, notice and comment rulemaking is that an agency’s action may be set aside if found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law,” 5 U.S.C. § 706(2)(A). The Supreme Court recently has reiterated a restrictive view of the arbitrary and capricious standard, stating:
“The scope of review under the ‘arbitrary and capricious’ standard is narrow and a court is not to substitute its judgment for that of the agency. Nevertheless, the agency must examine the relevant data and articulate a satisfactory explanation for its action including a ‘rational connection between the facts found and the choice made.’ ... In reviewing that explanation, we must ‘consider whether the decision was based on a consideration of the relevant factors and whether there has been a clear error of judgment.’ ... Normally, an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be as[626]*626cribed to a difference in view or the product of agency expertise.”
Motor Vehicle Manufacturers Ass’n v. State Farm Mutual Automobile Insurance Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 2866-67, 77 L.Ed.2d 443 (1983) (citations omitted). In addition, as part of our review, we also determine: (1) whether the agency acted within the scope of its authority, and (2) whether the agency complied with the prescribed administrative procedures. See Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 415-17, 91 S.Ct. 814, 823-24, 28 L.Ed.2d 136 (1971); American Petroleum Institute v. EPA, 540 F.2d 1023, 1028 (10th Cir.1976), cert. denied, 430 U.S. 922, 97 S.Ct. 1340, 51 L.Ed.2d 601 (1977); 5 U.S.C. § 706(2)(A) (C) &(D).
II
A preliminary question arises as to what documents we may review in considering these petitions. The EPA, supported by the environmental petitioners, has moved to strike references in the briefs filed by industry petitioners to documents and reports not in the record. Industry petitioners not only allege that those items are proper for our consideration but have moved to supplement the record to include the documents and reports that they cite. A decision on the appropriate use of these materials is important, because, in this ease, we are a reviewing body, not an independent decision maker. We do not substitute our judgment for the judgment of the agency simply because we might have decided matters differently. We agree with the comments in Deukmejian v. Nuclear Regulatory Commission, 751 F.2d 1287, 1323-1326 (D.C.Cir.1984), that the agency’s action must be reviewed on the basis articulated by the agency and on the evidence and proceedings before the agency at the time it acted. Aggressive use of extra-record materials also would run directly counter to the admonitions of the Supreme Court in Motor Vehicle Manufacturers Association.
Thus, any exception to this general rule against the use of extra-record materials must be extremely limited. Nevertheless, a few courts have found exceptions. A recent law review article discusses the problem that we, and all other appellate courts, face in determining whether and how to use extra-record citations. Stark & Wall, Setting No Records: The Failed Attempts to Limit the Record in Review of Administrative Action, 36 Ad.L.Rev. 333, 335 (1984). The article notes that, on review, parties have offered extra-record studies and other evidence under a number of justifications, including: (1) that the agency action is not adequately explained and cannot be reviewed properly without considering the cited materials, see Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 420, 91 S.Ct. 814, 825, 28 L.Ed.2d 136 (1971); (2) that the record is deficient because the agency ignored relevant factors it should have considered in making its decision, see Hiatt Grain & Feed, Inc. v. Bergland, 446 F.Supp. 457, 467 (D.Kan.1978), aff'd on other grounds, 602 F.2d 929 (10th Cir.1979), cert. denied, 444 U.S. 1073, 100 S.Ct. 1019, 62 L.Ed.2d 755 (1980); (3) that the agency considered factors that were left out of the formal record, see Environmental Defense Fund, Inc. v. Blum, 458 F.Supp. 650, 661 (D.D.C.1978); (4) that the case is so complex and the record so unclear that the reviewing court needs more evidence to enable it to understand the issues, see Bunker Hill Co. v. EPA, 572 F.2d 1286, 1292 (9th Cir.1977); and (5) that evidence coming into existence after the agency acted demonstrates that the actions were right or wrong, see American Petroleum Institute v. EPA, 540 F.2d 1023,1034 (10th Cir.1976), cert. denied, 430 U.S. 922, 97 S.Ct. 1340, 51 L.Ed.2d 601 (1977). As Stark and Wall observe, when such justifications are offered the court is forced as a practical matter to examine the material, whether or not motions to supplement the record are granted. 36 Ad.L.Rev. at 343-44.
In the instant case one or more of the above justifications are advanced with respect to virtually all extra-record cita[627]*627tions. We have in fact considered those materials to see if they fall within any of these possible justifications. But the fact that we examine these materials for the limited purposes set forth above does not mean that the cited items should become a part of the record in these cases. The references to such extra-record items, we believe, are substantially akin to the practice of citation to scientific treatises in ordinary civil cases. Therefore, we deny the EPA’s motion to strike the references; but we also deny all motions to supplement the record except the EPA’s unopposed motion to supplement the record with a letter of transmittal dated June 8, 1983, from Christopher Herman of the EPA to Larry Boggs, counsel for petitioner American Mining Congress and two accompanying memoranda dated April 27, 1983, and June 7, 1983.
Ill
The industry petitioners6 contend that the language and legislative history of the UMTRCA require the EPA to find that the mill tailings piles pose a “significant risk” before the EPA may promulgate regulations dealing with the mill tailings piles. Petitioners rely on Industrial Union Department, AFL-CIO v. American Petroleum Institute, 448 U.S. 607, 100 S.Ct. 2844, 65 L.Ed.2d 1010 (1980) (hereinafter the Benzene case). In the Benzene case, Justice Stevens, for the plurality, invalidated OSHA regulations intended to protect workers from the carcinogenic effects of benzene. Justice Stevens and three other members of the Court agreed that under §§ 3(8) and 6(b)(5) of the Occupational Safety and Health Act,7 before the Secretary may promulgate any permanent standards for workers’ health and safety, he must find that their workplace presents a significant risk of harm. Id. at 642, 100 S.Ct. at 2864. In so concluding, the plurality relied heavily upon the actual language from §§ 3(8) and 6(b)(5). Id. at 641-42, 100 S.Ct. at 2863-64.
We believe that the Benzene case and all the other OSHA cases requiring a threshold finding of significant risk are readily distinguishable from the case at hand. In the UMTRCA Congress commanded the EPA, the NRC, and the DOE to deal with the problems posed by uranium mill tailings:
“The Congress finds that uranium mill tailings located at active and inactive mill operations may pose a potential and significant radiation health hazard to the public, and that the protection of the public health, safety, and welfare and the regulation of interstaté commerce require that every reasonable effort be made to provide for the stabilization, disposal, and control in a safe and environmentally sound manner of such tailings in order to prevent or minimize radon diffusion into the environment and to prevent or minimize other environmental hazards from such tailings.”
42 U.S.C. § 7901(a) (emphasis added). It would be disingenuous to hold, after reading Congress’ own statement of its findings and purposes, that the EPA must make its own determination of whether radon emissions present a risk significant enough to warrant regulation under the UMTRCA. Admittedly, the language “may pose a potential ánd significant radiation health hazard to the public” might lead one to question whether Congress was resolute on the degree of danger the mill tailings pose. Yet, even if Congress was unsure of the absolute risk posed by mill tailings, it was sure of its desire to stabilize and dispose of the tailings and that “every reasonable effort be made to provide for the stabilization, disposal, and control in a safe and environmentally sound man-ner____” Id.
[628]*628The legislative history of both the UMTRCA and its 1982 amendment confirm our belief that it is unnecessary for the EPA to make its own determination that the tailings pose a significant risk. The House Report on the UMTRCA stated, “[a]s a result of being for all practical purposes, a perpetual hazard, uranium mill tailings present the major threat of the nuclear fuel cycle.” H.R.Rep. No. 1480, Part 1, 95th Cong., 2d Sess. 11, reprinted in 1978 U.S.Code Cong. & Ad.News 7433, 7433. Thus, Congress considered the mill tailings a hazard, despite evidence brought before it that mill tailings might not pose an immediate danger to current generations. In part two of the same House Report the NRC Chairman, Dr. Joseph M. Hendrie, described how the mill tailings piles present a hazard to the public health:
“Unlike high-level radioactive waste from the back end of the nuclear fuel cycle, which contains products of the fission reaction, mill tailings contain only naturally occurring radioactive elements, in small quantities____
The health effects of this radon production are tiny as applied to any one generation, but the sum of these exposures can be made large by counting far into the future, large enough in fact to be the dominant radiation exposure from the nuclear fuel cycle.”
Id. at 25, 1978 U.S.Code Cong. & Ad.News at 7433, 7452. By including such passages in its report, Congress showed it was clearly aware that the mill tailings themselves did not pose an immediate grave danger to persons in being. Nevertheless, Congress chose to consider protecting future generations by enacting the UMTRCA and requiring the immediate stabilization and disposal of those tailings.
The 1982 amendments did not change the thrust of the original UMTRCA plan. The changes in the legislation that pertained to the EPA’s duties under the UMTRCA, according to the House Conference Report, were not designed to change the EPA’s basic regulatory approach. House Conf. Rep. No. 884, supra, at 47, 1982 U.S.Code Cong. & Ad.News at 3592, 3617. Instead, Congress designed the amendments to set out the factors that the EPA should consider in developing the standards. The Conference Report indicated that Congress did not find fault with the EPA’s proposed standards.
“In each instance, the conferees have agreed to include specific references in the appropriate sections of the Atomic Energy Act directing EPA and NRC, in promulgating such standards or regulations, to consider the risk to the public health, safety, and the environment, the environmental and economic costs of such standards of [sic] regulations, and such other factors as EPA or NRC, respectively, determine to be appropriate.
... The conferees note that this language reflects accurately the current regulatory approach of the agencies. The language agreed to by the conferees should not result in any delays in establishment of remedial action standards. EPA, for example, has already advised the conferees that it is considering costs in formulating its inactive site requirements---- Moreover, in adopting the language, the conferees intend neither to divert EPA and NRC from their principal focus on protecting the public health and safety nor to require that the agencies engage in cost-benefit analysis or optimization.
■ The conferees are of the view that the economic and environmental costs associated with standards and requirements established by the agencies should bear a reasonable relationship to the benefits expected to be derived. This recognition is consistent with the accepted approach to establishing radiation protection standards, and reflects the view of the conferees that, in promulgating such general environmental standards and regulations, EPA and NRC should exercise their best independent technical judgment in making such a determination. At all times, the conferees fully intend [629]*629that EPA and NRC recognize as their paramount responsibility protection of the public health and safety and the environment.”
Id.
In this conference report there was no mention of the necessity of finding “significant risk” before promulgating regulations. We believe it would be outside the appropriate bounds of judicial review for us to require a showing of significant risk when Congress has been clear in its approval of the EPA’s approach to mill tailing regulation.8
IV
The industry petitioners argue that the EPA’s radon “flux” or emission limits and radium-in-soil cleanup standards exceed the EPA’s statutory authorization because they operate inside the mill tailings sites. This argument that the EPA may not adopt standards that operate at the mill tailings sites derives from the 1970 Reorganization Plan which transferred the Atomic Energy Commission (now NRC) authority to set generally applicable environment standards to the EPA. See generally Ouivira Mining Co. v. EPA, 728 F.2d 477, 480 (10th Cir.1984) (history of reorganization). That [630]*630act defined standards to mean “limits on radiation exposures or levels, or concentrations or quantities of radioactive materials, in the general environment outside the boundaries of locations under the control of persons possessing or using radioactive material.” 84 Stat. 2086, 2088 (1970) (codified at 5 U.S.C.App. § 2(a)(6) at 1132-33). The AEC (now NRC) was to retain authority over the licensing of commercial operations using or producing radioactive materials. 5 U.S.C.App. at 1135.
The American Mining Congress argues that this strict distinction between the EPA operating outside site boundaries and the NRC operating on-site has been maintained in the UMTRCA. The reference to “outside the boundaries,” however, does not appear in the final version of § 275 of the UMTRCA. 42 U.S.C. § 2022(b). As finally enacted that statute requires the EPA to adopt only “standards of general application,” leaving “implementation and enforcement” to the NRC in the conduct of its licensing activities. Id. at § 2022(d). The legislative history and administrative interpretations prior to the promulgation of these regulations are less than clear. Yet Congress’ concern all along has been the protection of the general public from radiation and other hazards connected with these operations. See House Conf.Rep. No. 884, supra, at 47, 1982 U.S.Code Cong. & Ad.News at 3617. If one of the principal hazards is inert radon gas that escapes from a pile into the atmosphere and travels some distance, the obvious remedy is to prevent that escape. Such remedy requires measures applicable at the source. We hold that the EPA has not exceeded its statutory authority in adopting this remedy.
V
The American Mining Congress argues that the radon “flux” or emission limits and radium-in-soil cleanup standards limits are invalid because they influence, and to a great extent define, the type of engineering or design standard to be selected by the implementing agency. See H.R.Rep. No. 1480, Part 1, supra, at 17, reprinted in 1978 U.S.Code Cong. & Ad.News at 7433, 7439 (“The EPA standards and criteria should not interject any detailed or site-specific requirements for management, technology or engineering methods ... on the Department of Energy.”).
As noted above, Congress sought in the UMTRCA to divide responsibility for disposal and cleanup among the agencies, with the EPA setting standards for general application and the DOE and the NRC implementing those standards. See 42 U.S.C. §§ 7911-7925. This division of responsibility and authority, however, should not be read to prevent the EPA from instituting a radon flux or radium limit that could be translated into a covering measurement requirement. As we discuss in the next section, the EPA had to meet a congressional mandate of some sort of cost-benefit analysis in promulgating the regulations. To do so it had to be able to estimate the cost of implementing the regulations. To determine the cost, the EPA had to have an approximate idea of the type of remedial action that will be undertaken; otherwise, its task is impossible. We therefore hold that the EPA did not exceed its authority by enacting emission limitations that appear to correlate directly with covering measurements. Furthermore, because the standards are general in nature — they apply to all sites — we do not view them as site-specific “management, technology or engineering” methods.
VI
All petitioners criticize the EPA’s approach in considering costs and benefits in developing the standards. The industry petitioners find fault with what they describe as the EPA’s failure to consider costs of disposal and cleanup in comparison with what they perceive as the limited health benefits of the regulation. The environmental petitioners,9 on the other hand, ar[631]*631gue that the EPA should promulgate strict feasibility standards, with a very limited review of the costs of implementing those standards.
The environmental petitioners argue that the UMTRCA requires the EPA to protect the public health to the maximum extent possible, constrained only by technical feasibility, and, to some extent economic feasibility. Feasibility analysis, as the Supreme Court has used that term, places a less severe restraint on agency action intended to forestall environmental harm than cost-benefit analysis. See American Textile Manufacturers Institute, Inc. v. Donovan, 452 U.S. 490, 507-09 & n. 27, 101 S.Ct. 2478, 2489-90 & n. 27, 69 L.Ed.2d 185 (1981). Feasibility analysis and cost-benefit analysis are mutually exclusive approaches. Id. at 509, 101 S.Ct. at 2490. In light of the language of the 1982 amendment to the UMTRCA we must reject the environmental petitioners’ argument that control of mill tailings should be as extensive as is capable of being performed. The amendment foreclosed the EPA’s use of a feasibility standard in promulgating regulations by requiring the EPA to consider “the environmental and economic costs” of applying the standards. 42 U.S.C. § 2022(a). The legislative history also reveals Congress’ intent to forgo a feasibility standard. The conference report noted that there should be a reasonable relationship between the environmental and economic costs associated with the standards and the benefits to be derived from them. House Conf.Rep. No. 884, supra, at 47, 1982 U.S.Cong. & Ad.News at 3617.10
We agree with the industry petitioners that the UMTRCA does require a consideration of costs relative to benefits, a cost-benefit analysis,11 by requiring a reasonable relationship between costs and benefits. We have no doubt, after reading the UMTRCA and its legislative history, that Congress gave the EPA a clear indication of its intent to require some sort of cost consideration. See American Textile Manufacturers, Inc., 452 U.S. at 510-11, 101 S.Ct. at 2491.
Nevertheless, although Congress acknowledged a cost-benefit requirement, it did not define what the EPA must do to meet that requirement in promulgating regulations. The label “cost-benefit analysis” encompasses everything from a strict mathematical balancing formula to a less strict standard that merely requires the agency to recognize both the costs and benefits of specific proposed alternatives and consider the differences in choosing an appropriate alternative. “Labels are neither important nor determinative.” American Petroleum Institute v. EPA, 540 F.2d 1023, 1037 (10th Cir.1976), cert. denied, 430 U.S. 922, 97 S.Ct. 1340, 51 L.Ed.2d 601 (1977).
[632]*632We read the UMTRCA to provide that the EPA must consider the costs involved in the regulations and, with the guidance of Congress' intent, find that these costs bear a reasonable relationship to the benefits derived. See House Conf. Rep. No. 884, supra, at 47, 1982 U.S.Code Cong. & Ad.News at 3617. The EPA must consider Congress’ intent to protect the public health and safety and the environment in its decision, because this intent refines the cost-benefit standard by assigning the values or weights that the EPA must use in its determination of a balance of benefits and costs. Id. (Congress considered protection of these things to be “paramount”); see generally American Petroleum Institute, 540 F.2d at 1028 (“the guiding star is the intent of Congress ____ All issues must be viewed in the light of that intent.”). The EPA labeled its final standard for control of the tailings piles as an “optimized cost-benefit” alternative. 48 Fed.Reg. at 591. The legislative history of the statute specifically stated that an optimized cost-benefit analysis was not necessary. It also stated that even a cost-benefit analysis was not required.
In the House Conference Report, discussing the 1982 amendment, it states,
“The language agreed to by the conferees should not result in any delays in establishment of remedial action standards. EPA, for example, has already advised the conferees that it is considering costs in formulating its inactive site requirements____ Moreover, in adopting the language, the conferees intend neither to divert EPA and NRC from their principal focus on protecting the public health and safety nor to require that the agencies engage in cost-benefit analysis or optimization. ”
House Conf. Rep. No. 884, supra, at 47, reprinted in 1982 U.S.Code Cong. & Ad. News at 3592, 3617 (emphasis added). The topic sentence of the next paragraph states, however, that “[t]he conferees are of the view that the economic and environmental costs associated with standards and requirements established by the agencies should bear a reasonable relationship to the benefits expected to be derived.” Id. This language, in the context of the entire legislative history of the 1982 amendments, see infra notes 10 and 11, convinces us that Congress intended cost-benefit analysis, but less strict than an optimized cost-benefit analysis.
VII
We must examine each standard to determine whether the EPA properly considered in its analysis each factor required by Congress. Moreover, we must determine whether the EPA’s record supports the EPA’s action. See Ethyl Corp. v. EPA, 541 F.2d 1, 36 (D.C.Cir.) (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976).
A
The industry petitioners criticize the EPA’s change in and justification for: its radon emission standard (allegedly from a risk of lung cancer to a risk of misuse); its linear-nonthreshold calculations of radon risk; and its projected life-savings from the promulgation of the regulations. They also criticize the insubstantial benefits of the regulations, which they say bear no rational relationship to their substantial costs.
We do not believe that misuse is an unreasonable justification for the standard, particularly in light of the fact that misuse of the tailings helped lead to the enactment of the UMTRCA. See H.R.Rep. No. 1480, Part 1, supra, at 11, 1978 U.S.Code Cong. & Ad.News at 7433, 7434.12 We do not [633]*633believe that the EPA shifted its justification for the radon standard in a manner that denied any party the opportunity to comment on the alleged newly asserted justification, misuse of tailings. In its description of the proposed standard for “Control of Tailing Piles” the EPA stated: “Many interrelated factors affect the long-term performance of tailings pile disposal methods. They include external natural phenomena, such as earthquakes, floods, windstorms, and glaciers, internal chemical and mechanical processes, and human activities." 46 Fed.Reg. at 2558 (emphasis added). Thus, given the legislative history of the Act and the EPA’s own language in the proposed standards, misuse, as a justification for the standards, was sufficiently foreshadowed in the proposed scheme; the parties here were not deprived of notice or an effective opportunity to respond. See Small Refiner Lead Phase-Down Task Force v. EPA, 705 F.2d 506, 547 (D.C.Cir.1983); Beirne v. Secretary of Department of Agriculture, 645 F.2d 862, 865 (10th Cir.1981).
This court recognizes the industry concerns over the costs that must be incurred to prevent, using the EPA’s calculations, approximately 170 to 240 potential lung cancer deaths per century in the United States from uncontrolled mill tailings from all inactive sites. 48 Fed.Reg. at 593; I FEIS-IN at 67. We think there is reasonable authority to support the EPA’s method of risk calculation in regard to its potential lung cancer deaths estimate.13 See Reserve Mining Co. v. EPA, 514 F.2d 492, 507 n. 20 (8th Cir.1975) (en banc) (“Indeed, a number of the disputes involve conflicting theories and experimental results, about which it would be judicially presumptuous to offer conclusive findings.”).
The industry petitioners also criticize the EPA’s “estimate that people living continuously next to some of the piles may have lifetime excess lung cancer risks as high as 4 chances in 100.” 48 Fed.Reg. at 593, 598. They criticize these figures on several bases,14 arguing that: (1) few people, if any, are subjected to this risk; (2) the estimate is unreasonable in its assumption that someone would live continuously near a tailings pile; (3) the radon daughter concentrations to which nearby individuals are allegedly exposed, which the EPA listed in the Final Environmental Impact Statement, do not correlate with the radon concentrations in the study the EPA cites; (4) the EPA based the radon measurements on inadequate sampling data; and (5) the risk measurements include background radiation, not just radon from the tailings piles. We address separately these arguments.
The EPA has made a “4 in 100” excess risk of lung cancer calculation for those residing near the inactive sites in its radon risk calculation.15 This figure defines both [634]*634the risk created by the inactive mill tailings and the benefit to be derived from instituting the radon emission standard — avoidance of that risk. We believe the evidence in the record reasonably supports this risk estimate figure.
Most of the inactive uranium mill tailing sites are located in rather sparsely populated areas. Congress no doubt was aware of this fact when it enacted the UMTRCA and designated the specific sites to which the regulations would apply.16 See 42 U.S.C. § 7912. There are, however, piles located near cities with sizeable local populations, such as Salt Lake City, Utah, and Grand Junction, Colorado. See I FEIS-IN at 60, Table 4-3. In addition, as the EPA points out, population distributions may change, as a consequence exposing more people to the dangers of radon. I FEIS-IN at 59 (“Although we have ignored population changes since 1970, a future increase in population at several of the urban sites seems likely.”). Therefore, although we may agree with the industry petitioners that few people are currently exposed to radon from the piles, this fact does not necessarily negate the need for regulation, especially in light of Congress’ presumed knowledge regarding the population density in areas surrounding the designated sites.17
We also reject industry petitioners’ assertion that it is unreasonable that someone would live continuously near a pile. Supposedly rational people in our society continuously expose themselves to risks of greater magnitude than those involved here: workers with nuclear materials, underground coal miners, and many others. Some do it for money, some because they underestimate the risk. Radon emission from these piles will occur for thousands of years unless prevented. Public awareness of the dangers may wane. The industry petitioners’ criticism does not undermine the validity of the risk assessment; it only points to the limited number of persons who currently may be subject to the risk.
Industry petitioners are correct that there were some discrepancies between the exposure level estimates used to calculate the risk factor included in the Final Environmental Impact Statement, and the referenced study conducted by consultants Ford, Bacon and Davis Utah, Inc.18 See Letter from Christopher Herman, Attorney, Air, Noise and Radiation Division, United States Environmental Protection Agency, Brief for American Mining Congress, Addendum F. Although these discrepancies alter the risk estimates, we do not think that the deviation is sufficiently significant to render the EPA’s upper limit risk estimate to be unsupported by available data. Precision is desirable, yet these discrepancies do not rise to a level that would require invalidation of an estimate as an arbitrary decision.
Industry petitioners criticize the EPA for calculating risk from figures they allege were derived from inadequate sampling [635]*635data and data that included background levels of radiation, instead of figures based solely on the radon emitted from the inactive sites. We do not find these flaws fatal to the EPA’s conclusions. The sampling data compiled by the researchers for the EPA is adequate to support the EPA’s estimates.19 An estimate of risk may reasonably include possible background risk— to exclude it could lead to unrealistic assumptions regarding the dangers that the sites do present.
B
The industry petitioners’ arguments against the EPA’s radium-in-soil concentration standard, 40 C.F.R. § 192.12(a) (1984), are similar. They say the standards are invalid because (1) the EPA’s assumptions about the likelihood of significant exposures from the concentrations do not support the standard; and (2) the EPA’s assumptions about the relationship between radium concentrations in the soil and indoor radon exposure and potential health effects are arbitrary.
As the EPA points out, contaminated soil under or near a building contributes significantly to indoor radon. 48 Fed. Reg. at 600. The EPA admits in the FEIS-IN, and the industry petitioners point out, that the actual exposure to radon from contaminated soil may vary from the levels that the EPA assumed. Even considering the possible inaccuracies of the EPA’s assumptions, there is nothing in the record or arguments presented to convince us that the EPA acted arbitrarily or capriciously in promulgating this standard to deal with a significant danger the tailings piles present.
The State of Colorado claims that the EPA provided inadequate notice and opportunity for comment on the one hundred square meter sample area for the radium concentration standard. Although this sort of area designation admittedly could influence the effectiveness of the standard, it is not the sort of change that necessitates a new round of comment.
C
In its final standard the EPA increased the allowable level of indoor radon concentration from 0.015 WL to a maximum level of 0.03 WL, stressing that reasonable ef[636]*636fort should be made to achieve a radon concentration level of 0.02 WL. See 40 C.F.R. § 192.12(b)(1) (1984). The EPA justified this change on the basis that it would ease implementation of the final standard. See 48 Fed.Reg. at 600. It further stated that “the final standard deals adequately with complications introduced by the presence of any high concentration of naturally-occurring radionuclides, and avoids unnecessary and costly remedial actions that produce only marginal improvements.”20 Id.
The environmental petitioners criticize the final standard for allowing an unjustified increase in risk, represented by the changes from the proposed to the final standard. The proposed standard of 0.015 WL had an estimated residual risk of lung cancer of 0.8 in 100. I FEIS-IN at 108, Table 7-1. The “reasonable effort” level of the final standard, 0.02 WL, had an estimated residual risk of lung cancer of 1.3 in 100. Id. Contrary to the environmental petitioners’ argument, the EPA did evaluate the residual risk of lung cancer of the ceiling level of radon decay product, 0.03 WL including background, by considering a range of levels between 0.0 WL above background and 0.05 WL above background. Id. The FEIS-IN shows a residual risk of lung cancer of 5 in 100, which appears to be the top side risk at 0.05 WL above background. Although this is a significant increase in the risk, we cannot say the EPA acted arbitrarily or capriciously or beyond its authority in allowing the higher level and its attendant risk.
D
Accepting the EPA’s risk calculation estimates and thus its conclusion that the control of the tailings piles will reduce health risks and provide benefits Congress desired, we must now determine whether there existed a reasonable relationship between the estimated benefits of control of the tailings piles and the costs of providing those benefits. We start with the obvious premise that there can be no reasonable relationship in the abstract. As we indicated earlier, Congress’ intent serves to define the reasonableness of any relationship between costs and benefits. Given Congress’ admonition that the EPA recognize as its “paramount responsibility protection of the public health and safety and the environment,” House Conf.Rep. No. 884, supra, at 47, 1982 U.S.Code Cong. & Ad.News at 3592, 3617, we review the EPA’s adoption of the 20 pCi/m2/sec radon emission standard.
In the Final Environmental Impact Statement the EPA analyzed the “Benefits Derived From Controlling Uranium Mill Tailings Piles,” specifically considering the benefits that would be provided under various remedial alternatives — ranging from “No Standards” to the EPA’s originally “Proposed Standard.” See I FEIS-IN at 99, Table 6-6. The benefits enumerated were: the probable reduction or elimination of human misuse of the tailings after control; the permanence of the controls inhibiting misuse and erosional spreading; the reduction in vulnerability of the site to flooding; the reduction in the residual risk of lung cancer; the deaths avoided; and the surface water protection provided. The alternative the EPA ultimately selected in the final regulations, a 20 pCi/m2/sec standard, [637]*63721 provides significant benefits, meeting Congress’ concern that the standard protect the public health, safety, and environment. The selected standard and its corresponding control method,22 see I FEIS-IN at 91, according to the Final Environmental Impact Statement, will most likely prevent misuse of the tailings for over a thousand years, prevent erosional spreading for many thousands of years, leave no sites vulnerable to flooding, reduce residual risk of lung cancer by over 95%, avoid 190 deaths in the first 100 years of protection, and protect surface water from contamination for many thousands of years. I FEIS-IN at 99, Table 6-6.
The environmental petitioners criticize the validity of the final radon emission standard for not “preventing or minimizing” diffusion of radon into the atmosphere. See 42 U.S.C. § 7901(a) & (b)(1). In the same vein, they also criticize the tenfold increase in the level of emissions permitted by the final standard, an increase from 2 pCi/m2s to 20 pCi/m2s. Despite this significant increase, we believe that the EPA, through its final radon standard, has sought to prevent or minimize radon emission, thus meeting the Congress’ mandate. See I FEIS-IN at 99, Table 6-6, at 101. Given the reduction in risk provided by the final standard, we do not believe that the “tenfold increase” in the level of emissions allowed by the final standard makes that standard unreasonable, even in light of the change. We hold that the final standard is within a zone of reasonableness, see Small Refiner Lead Phase-Down Task Force, 705 F.2d at 525, and a logical outgrowth of the proposed standard, id. at 547.
All of the petitioners find fault with the EPA’s suggested longevity requirement for the final radon emission standard — that the standard control emissions to the extent reasonably achievable, have an effective life of 1000 years, and in any case, for at least 200 years. 40 C.F.R. § 192.02(a) (1984). The industry petitioners consider the thousand-year longevity goal unrealistic. The environmental petitioners consider the reduced two hundred-year minimum inadequate, especially in light of the thousand-year proposed standard. We cannot say the final standard fails to accomplish Congress’ goal of long-term stabilization in view of Congress’ command to take costs into consideration. The final standard meets Congress’ desire to avoid the health risks presented by the tailings piles and provide a remedial program that will result in long-term stabilization of the piles. .
Congress stressed that the “remedial action must be done right the first time.” H.R.Rep. No. 1480, supra, at 40, 1978 U.8. Code Cong. & Ad.News at 7433, 7467. Given Congress’ expressed desire to have a remedial program that will stabilize the tailings piles and additionally, not require continuous allocations of funds, id., we believe the EPA acted reasonably in promulgating a standard based primarily upon the adoption of a passive measure, an earth covering. The industry petitioners particularly are critical of EPA’s refusal to consider remedial measures that are more dependent upon active controls such as “maintenance” and “monitoring” by the implementing agencies. The EPA did consider such an alternative. I FEIS-IN at 102, Alternative E. But it rejected that alternative upon review of the benefits it would provide. Id. at 98-104, 128-29. Arguably, EPA could have placed more reliance upon the maintenance and monitoring activities of the implementing agencies. Congress, however, did not require it to do so and instead indicated its preference for a more permanent remedial program. The EPA’s understanding of its duty under the statute to rely principally upon passive control is sufficiently rational to withstand our review. See Chemical Manufacturers Association v. National Resources Defense [638]*638Council, Inc., — U.S.-,-, 105 S.Ct. 1102, 1107, 84 L.Ed.2d 90 (1985).
EPA’s cost estimates for remedial action at the mill sites and for off-site cleanup were 158 and 38 million (1981) dollars, respectively. EPA stated that the DOE had estimated its additional costs for program development and management (“overhead”) at 118 million (1981) dollars. The total estimated expenditure was , 314 million (1981) dollars. These figures include both cleanup (off-site) and disposal (on-site) costs. 48 Fed.Reg. at 596. We recognize the tremendous costs that this cleanup program represents. Nevertheless, we cannot condemn as arbitrary the EPA’s determination that these costs bear a reasonable relationship to the benefits to be provided. In fact, in enacting the UMTRCA Congress was well aware of the potential costs of the program. Assessing the economic impact of the legislation in 1978, the House Report stated:
“This legislation is not expected to have any significant inflationary impact. Over the next 7 years, 22 tailing sites will be treated at a total cost ranging anywhere from $15 million to $200 million, depending largely upon whether tailings will be treated and stored at their present location or, instead, moved to newly prepared disposal sites. Little of this cost is expected to be incurred during the next 3 years because of the time required to identify and prepare disposal sites. Additional costs may be borne by individual states if new disposal sites are required. But even taking these additional costs into account, the impact of the legislation on inflation and overly [sic] economic performance is expected to be immeasurable [sic].”
H.R.Rep. No. 1480, supra, at 47, 1978 U.S. Code Cong. & Ad.News at 7433, 7474. We must defer to the EPA’s finding of a reasonable relationship between the costs and the benefits of the remedial action program for disposal of the mill tailings at the inactive sites. As we have said before in another context, “[w]e are convinced that EPA made a serious, careful, and comprehensive study of the costs which compliance will impose____ If Congress believes that the cost is too high, it can amend the Act. All we say is that EPA has complied with the statutory mandate.” American Petroleum Institute v. EPA, 540 F.2d 1023,1038 (10th Cir.1976), cert. denied, 430 U.S. 922, 97 S.Ct. 1340, 51 L.Ed.2d 601 (1977).
VIII
There is one area in which we agree with some of the petitioners or the intervenor that the EPA standards are not in compliance with the law or are inadequate and require remand.
In its final standards the EPA abandoned its proposed general standards for limits of concentrations of toxic substances in surface and ground water.23 See 48 Fed.Reg. at 591, 594; see also 46 Fed.Reg. at 2562 (text of proposed standards). The EPA concluded that potential contamination of surface and ground water should be dealt with on a site-specific basis. 40 C.F.R. § 192.20(a)(2) (1984); see also 48 Fed.Reg. at 594 (discussion of final rules). In place of the proposed standard for control of waterborne pollutants the EPA included a guideline, stating that the DOE should assess each site individually and establish any corrective or preventive programs necessary to meet relevant state and federal water quality standards. 48 Fed.Reg. at 591. It stated that those programs should be consistent, to the maximum extent practicable, with the Solid Waste Disposal Act (SWDA), 42 U.S.C. §§ 6901-6986. 48 Fed. Reg. at 591; 40 C.F.R. § 192.20(a)(2) & (3) (1984).
We agree with the environmental petitioners and the State of Colorado that by its actions the EPA violated its duty under the statute. In its proposed regulations the EPA stated that there was a problem with water contamination: “There is evidence of limited ground water contamination at some of the inactive sites, but [639]*639the prospects for long-term contamination have not been fully assessed.” 46 Fed. Reg. at 2560. It then proposed specific standards for water quality generally. Id. at 2559-63. In the final regulations the EPA acknowledges there are problems, 48 Fed.Reg. at 593, but states: “We do not believe that the existing evidence indicates that ground water contamination from inactive mill tailings is or will be a matter of regulatory concern.” Id. at 599. We do not believe that the EPA, in issuing the final regulations, declared that it was wrong before; indeed, by stating that potential contamination should be dealt with on a site specific basis, id. at 594, it acknowledged that problems exist at some sites.
The draft Environmental Impact Statement gave little indication of the possibility that the EPA would reject the proposed water standards and adopt a “guidance” outline for the implementing agencies, abandoning any attempt at specific numerical limits of toxic elements. The only statement we could find in the draft EIS that could be construed to indicate consideration of an alternative similar to that adopted is this: “The proposed ground water protection standards could be considered too strict if implementing them would be unreasonably costly or if they would be impossible to apply.” Environmental Protection Agency, Draft Environmental Impact Statement for Remedial Action Standards for Inactive Uranium Processing Sites 8-8 (1980). The explanation of the proposed water quality standard anticipated possible increases in disposal costs to insure against future ground water contamination, but gave no indication that this possibility would make the proposed standards infeasible. 46 Fed.Reg. at 2560. Many of the comments submitted to the EPA during the comment period after publication of the proposed rules suggested that the EPA abandon its proposed ground water standards. See, e.g., II FEIS-IN at D-36, -37, -44, -45, -46. But, regardless of the tenor of some of the comments, the final “guidance” standard does not represent a logical outgrowth from the proposed regulations.
In our view when the EPA acknowledged that ground water contamination is a problem at some of the inactive sites, it was required to adopt general standards.24 The UMTRCA provides that the EPA “promulgate standards of general application.” 42 U.S.C. § 2022(a). The EPA was not to adopt “site-specific requirements.” H.R.Rep. No. 1480, Part I, supra, at 16-17, reprinted in 1978 U.S.Code Cong, and Ad.News at 7433, 7439. By directing the DOE to follow “relevant” state and federal quality standards consistent “to the maximum extent practicable” with the SWDA, see 48 Fed.Reg. at 591; 40 C.F.R. § 192.20(a)(2) and (3) (1984), the EPA was unlawfully delegating to the states or to the DOE its own rulemaking authority or was itself establishing standards that could vary from site to site.
We do not agree, however, with Colorado’s argument that an unlawful delegation is involved by the provision of the regulations allowing exceptions from the standards in particular circumstances — permitting the implementing agencies, the NRC and the DOE, to “select and perform remedial actions that come as close to meeting the otherwise applicable standard as is reasonable under the circumstances.” 40 C.F.R. § 192.22(a) (1984). So long as general standards are in place, permitting exceptions for special circumstances in which costs might be outrageous or compliance impossible is not an abrogation of the duty to promulgate general standards. A court could determine, if required to do so, the reasonableness of an exception granted in a specific case.
The environmental petitioners also complain that the EPA has a duty to for-[640]*640muíate general standards for nonradiological hazards, toxic chemicals in the uranium tailings, see 42 U.S.C. § 2022(a), but has not specifically addressed the subject except in conclusory form:
“We have reviewed the available data on toxic elements in tailings and improved the FEIS-IN in this respect [Appendix C]. We have concluded that it is reasonable to expect that hazards from toxic elements will be adequately limited if control and cleanup are carried out according to these final standards.”
48 Fed.Reg. at 597. We are satisfied that the EPA’s general regulations on radon emissions, requiring control of the movement of and probable covering of the tail-ings adequately deal with the nonradiological hazards of the toxic chemicals, except as they may enter waterways or underground water supplies. If there is a possibility that toxic chemicals will enter underground water supplies or waterways, the statute mandates that the problem be dealt with by general standards. On remand, the EPA will have to treat these toxic chemicals that pose a ground water risk as it did in the active mill site regulations.
IX
We reject all challenges to the regulations except as discussed in Part VIII above. The following regulation, concerning water contamination is set aside: 40 C.F.R. § 192.20(a)(2) — (8) (1984). The case is remanded to the agency for further consideration of that specific provision.
These exposure levels are in turn used to calculate absolute and relative risk estimates.