dissenting:
In Greater Boston Television Corp. v. FCC1 the late Judge Leventhal first enunciated what has come to be known as the “hard look” doctrine for judicial review of agency action.2 When today’s case first came to this court (Vermont Yankee I), Judge Bazelon upped the ante by endorsing the notion of a “good, hard look.”3 Now, notwithstanding clear directions to the contrary from the Supreme Court (in Vermont Yankee II),4 Judge Bazelon has applied a standard which is best described as requiring “too hard a look.” If there was ever a doubt prior to today, it is now clear that this court is committed to an assumed role as high public protector of all that is good from perceived evils of the nuclear age.
In his lengthy opinion today (Vermont Yankee III) Judge Bazelon adapts to a new heading an approach unanimously rejected by the Supreme Court in Vermont Yankee II and steps boldly into the arena of nuclear regulation to vacate all of the NRC's original, interim, and final fuel cycle rules — the product of nearly a decade of agency and judicial proceedings and an administrative record of thousands of pages in length. He justifies this action on a conclusion that the Commission somehow violated NEPA and the APA by adopting a generic table embodying a prediction that nuclear power plant wastes can someday be safely stored.
Judge Bazelon’s majority opinion has braided from NEPA, the APA, and his own perceptions of public good a novel standard of substantive and procedural scrutiny of NRC activity which we as a court are unauthorized to create and ill-equipped to manage. It has ignored the Supreme Court’s unequivocal mandate that we limit our examination in this matter to the administrative record to determine only whether the NRC’s rule was “arbitrary and capricious within the meaning of § 10(e) of the Administrative Procedure Act, 5 U.S.C. § 706.”5 On the record before us, which has been supplemented extensively since Vermont Yankee II, I cannot say that under the appropriate standard the rule fails. Accordingly, I would affirm.6
I. ANALYSIS
The cases before us today represent the latest chapter in perhaps the most celebrated administrative litigation of our time. Although the procedural backdrop of the litigation, adequately rehearsed in Judge Bazelon’s opinion,7 is extraordinarily complex, I believe the analysis and resolution of the central issues are comparatively simple.
That analysis must begin with a review of the case history, culminating in the unanimous directive given us by the Supreme Court in Vermont Yankee II. Both the holdings and the language of that opinion, essentially endorsing Judge Tamm’s original concurrence in Vermont Yankee I,8 clarify the appropriate standard for our review of the informal rulemaking challenged here. When we examine the agency’s rule-making under that standard, two further *68points become clear. First, the agency acted well within the scope of its broad authority by choosing to consider the issue of long-term risks and uncertainties from radioactive waste disposal genetically, rather than in a series of repetitive individual power plant licensings. Second, contrary to the majority’s joint assertions, the agency did not act arbitrarily and capriciously or “not in accordance with law” by concluding that licensing boards need not factor uncertainties into their environmental analyses, but, that “for the limited purposes of the fuel cycle rule it is reasonable to base impacts on the assumption which the Commission believes the probabilities favor”: namely, that sites can be found and the technology developed effectively to isolate radioactive waste from the biosphere.9
These conclusions are not only compelled by a careful review of the extensively supplemented record on appeal; they are totally consistent with all of the evidence reviewed in Judge Bazelon’s opinion. That record clearly reveals that the Commission has fulfilled its procedural obligations under § 102(2)(C) of NEPA 10 and the Administrative Procedure Act,11 has taken- the requisite “hard look” at the waste storage question, and has supported its final rule by “reasoned decisionmaking.” This court may require no more.
We easily resolve this appeal if we recognize the obvious — we live in a world of risk, uncertainty, and rapid technological change. We not only expect, we demand, expert agencies charged with broad statutory discretion to make predictions en route to factual, technical, and policy judgments. “Where existing methodology or research in a new area of regulation is deficient, the agency necessarily enjoys broad discretion to attempt to formulate a solution to the best of its ability on the basis of available information.” 12 So long as the agency adequately observes statutory procedures, gives close scrutiny to all relevant considerations, and engages in reasoned decision-making, it is not our job to substitute our judgment for the agency’s to reach what we perceive to be the best or correct result. Yet this is precisely what the majority has accomplished in this case.
II. BACKGROUND
A. The Challenged Rules
The rules at issue here13 resulted from an informal, “generic” rulemaking instituted by the NRC in November 1972 under 5 U.S.C. § 553 of the Administrative Procedure Act.14 This proceeding, like a number of generic rulemakings noticed by the Commission since 1972,15 was intended primarily to provide information for future adjudications in licensing proceedings. The product of the rulemaking was a table of numerical values (Table S-3), representing the incremental environmental effects associated with the uranium fuel cycle of one addition*69al light water power reactor.16 Individual licensing boards, in turn, were allowed to factor those representative values into the cost-benefit analysis for each proposed facility as part of the detailed Environmental Impact Statement (EIS) required by § 102(2)(C) of NEPA.17
Like all generic rulemakings, the proceeding here was instituted to further the goals of administrative efficiency, public participation, and uniformity in agency adjudications.18 The rulemaking permitted the agency to consider and settle in one proceeding a number of issues of fact or policy common to a large number of licensing hearings. Without the informal rulemaking, “generic” issues would have been reopened at each proceeding before individual licensing boards. This not only would have required, in proceeding after proceeding, an inefficient presentation of identical and repetitive evidence bearing on matters not truly “local” to the plant proposed for license,19 but likely would have resulted in inconsistent adjudications.
B. Vermont Yankee I
When this case first came to our court, environmentalist petitioners challenged Table S-3 on the grounds that its values “ex-presse[d] in numerical terms the conclusion that the environmental effects of the fuel cycle, including [nuclear power] waste disposal, are insubstantial.”20 Judge Bazelon’s opinion for the court set aside those portions of the rule pertaining to the issues of high-level waste disposal and waste reprocessing.21 At the outset the panel majority held, under § 102(2)(C) of NEPA, that “absent effective generic proceedings to consider [long-term waste disposal] issues, they must be dealt with in individual licensing proceedings.” 22 Judge Bazelon’s opinion for the panel then went on to find the agency’s generic proceedings “ineffective” for two interrelated reasons: because the agency’s procedures were inadequate to “ventilate” thoroughly the waste fuel disposal issues, and because the agency’s record was inadequate to support its rule.23 *70Thus, the issues were relegated to individual adjudication.
Judge Tamm, concurring in the result, joined only the second of the court’s two grounds, finding that “the inadequacy of the record demands that we remand this case to the Commission in order to ensure that it has taken a hard look at the waste storage issue.”24 Challenging Judge Bazelon’s first ground of decision, Judge Tamm noted that “the deficiency is not with the type of proceeding below, but with the completeness of the record generated.”25 He concluded,
The appropriate remedy at this point is not to impose ad hoc procedural requirements in an attempt to raise the level of petitioners’ participation, already adequate under [5 U.S.G.] section 553, but to remand for ... the documentation which the majority finds so conspicuously lacking. The Commission should be able to supply the court with a statement of the methods by which its staff arrived at the figures embodied in Table S-3 and by which [the NRC Waste Management Division Director] concluded that the waste storage problem is already technologically and economically soluble. If it cannot, we will have no choice but to invalidate the Commission’s rule under the “arbitrary, capricious” standard; if it can, we should defer to the administrative weighing of risks and benefits of additional reactors.26
C. Vermont Yankee II
The Supreme Court unanimously reversed the panel opinion.27 The united Court, speaking through Justice Rehnquist, resoundingly rejected the first of the panel’s decisional grounds:
[Njothing in the APA, NEPA, the circumstances of this case, the nature of the issues being considered, past agency practice, or the statutory mandate under which the Commission operates permitted the [circuit] court to review and overturn the rulemaking proceeding on the basis of the procedural devices employed (or not employed) by the Commission so long as the Commission employed at least the statutory minima, a matter about which there is no doubt in this case.28
The Court did not, however, reverse this court’s second finding, concurred in by Judge Tamm. Instead, it remanded that question to us with the following directive:
There remains, of course, the question of whether the challenged rule finds sufficient justification in the administrative proceedings that it should be upheld by the reviewing court. Judge Tamm, concurring in the result reached by the majority of the Court of Appeals, thought that it did not. There are also intimations in the majority opinion which suggest that the judges who joined it likewise may have thought the administrative proceedings an insufficient basis upon which to predicate the rule in ques*71tion. We accordingly remand so that the Court of Appeals may review the rule as the Administrative Procedure Act provides. We have made it abundantly clear before that when there is a contemporaneous explanation of the agency decision, the validity of that action must “stand or fall on the propriety of that finding, judged, of course, by the appropriate standard of review. If that finding is not sustainable on the administrative record made, then the Comptroller’s decision must be vacated and the matter remanded to him for further consideration.” Camp v. Pitts, 411 U.S. 138, 143 [93 S.Ct. 1241, 1244, 36 L.Ed.2d 106] (1973). See also SEC v. Chenery Corp., 318 U.S. 80 [63 S.Ct. 454, 87 L.Ed. 626] (1943). The court should engage in this kind of review and not stray beyond the judicial province to explore the procedural format or to impose upon the agency its own notion of which procedures are “best” or most likely to further some vague, undefined public good.29
The Vermont Yankee II Court buttressed its explicit mandate with four observations of major significance to our review on remand. First,- the Court reiterated the broad scope of NRC authority to deal with fuel cycle issues by rulemaking.30 Second, the Court declared that this court had “fundamentally misconceive[d] the nature of the standard for judicial review of an agency rule.”31
[Because] informal rulemaking need not be based solely on the transcript of a hearing held before an agency .. . the adequacy of the “record” in this type of proceeding is not correlated directly to the type of procedural devices employed, but rather turns on whether the agency has followed the statutory mandate of the Administrative Procedure Act or other relevant statutes32
The Court’s third observation, particularly important for our purposes today, emphasized the limited scope of our power to review NRC rulemaking under NEPA. Justice Rehnquist stated in no uncertain terms that a reviewing court’s already-limited power to review the substance of agency decisions is not significantly supplemented by the broadly stated environmental mandate found in § 101 of NEPA.33 In *72Calvert Cliffs’ Coordinating Comm. v. AEC,34 this court had held that § 101 imposes substantive restraints on agency actions, thus obliging courts to review agency decisions on the merits to ensure their consistency with the goals of NEPA.35 In Vermont Yankee II the Court acknowledged that while “NEPA does set forth significant substantive goals for the Nation, ... its mandate to the agencies is essentially procedural.”36 In Stryckers’ Bay Neighborhood Council v. Karlen,37 the Court elaborated on that statement:
Vermont Yankee cuts sharply against the Court of Appeals’ conclusion that an agency, in selecting a course of action, must elevate environmental concerns over other appropriate considerations. On the contrary, once an agency has made a decision subject to NEPA’s procedural requirements, the only role for a court is to insure that the agency has considered the environmental consequences ; it cannot “ ‘interject itself within the area of discretion of the executive as to the choice of the action to be taken.’ ”38
Furthermore, the Vermont Yankee II Court explicitly rejected Judge Bazelon’s view that NEPA authorizes courts to require agencies to “develop new procedures io accomplish the innovative task of implementing NEPA through rulemaking:”39
We have before observed that “NEPA does not repeal by implication any other statute.” ... In fact, just two Terms ago, we emphasized that the only procedural requirements imposed by NEPA are those stated in the plain language of the Act. . . . Thus, it is clear NEPA cannot serve as a basis for a substantial revision of the carefully constructed procedural specifications of the APA.40
Thus, after Vermont Yankee II, reviewing courts must not only limit their substantive review of agency action to the appropriate standards under the APA, they must take pains in their procedural review to avoid imposing requirements in excess of the “statutory minima” or extending their scrutiny beyond the “plain language” of § 102.
Vermont Yankee II’s fourth major pronouncement arose from its discussion of a consolidated case. Addressing Consumers Power Co. v. Aeschliman,41 which presented identical fuel cycle issues in the licensing context, the Vermont Yankee II Court reiterated both the breadth of NRC statutory authority and the limited “role of a court in reviewing the sufficiency of an agency’s *73consideration of environmental factors” under NEPA.42 The Court expressly recognized that § 102(2)(C) requires responsible agency officials to include in each EIS “a detailed statement . . . on . . . alternatives” to the proposed federal action43 and that NEPA obliges agencies to consider “every significant aspect of the environmental impact” of the action.44 Nevertheless, it refused to read those NEPA requirements as placing on the agency the burden of justifying every decision not to consider a highly unlikely alternative to a given action. The initial burden of establishing that an alternative is sufficiently material to require agency consideration, the Court suggested, properly rests with the challenger to the agency action.45 Furthermore, the Court reasoned, even when the challenger has met that burden, the agency’s subsequent duty to consider and explore little-known technological alternatives must necessarily be bounded by two practical constraints: the feasibility of those alternatives46 and the extent of information about them currently available.47
In short, the Vermont Yankee II Court’s discussion of Consumers Power confirmed a view this court has stated several times: that NEPA necessarily allows agencies some flexibility in determining the contents of their environmental impact statements in situations where uncertainty is abundant, technological alternatives are still developing, and much crucial information remains unavailable.48 In such circumstances, as *74Judge Wright noted in Scientists’ Institute for Public Information v. AEC,49 “NEPA is not a paper tiger, but neither is it a straitjacket.”50
Vermont Yankee II closed with the unambiguous command that courts should not employ judicial review of NRC action to encroach on fundamental policy choices in the nuclear power arena:'
Nuclear energy may some day be a cheap, safe source of power or it may not. But Congress has made a choice to at least try nuclear energy, establishing a reasonable review process in which courts are to play only a limited role. The fundamental policy questions appropriately resolved in Congress and in the state legislatures are not subject to re-examination in the federal courts under the guise of judicial review of agency action. Time may prove wrong the decision to develop nuclear energy, but it is Congress or the States within their appropriate agencies which must eventually make that judgment. In the meantime courts should perform their appointed function.51
III. OUR TASK ON REMAND
The majority has deviated from a clearly presented path — charted by the Supreme Court itself — to lead this court again on a confusing and unauthorized incursion into a policymaking area reserved by Congress to the NRC. I cannot participate in this trespass.
The Supreme Court’s explicit directive in Vermont Yankee II was that we first, clarify the nature of the agency decision under challenge; second, seek a “contemporaneous explanation ” of that decision; and third, judge the “propriety of that finding by the appropriate standard of review.”52 A conclusion that the agency’s finding was sustainable on the administrative record would effectively end our inquiry.
A. The Agency’s Decision
The primary NRC choice still under review is the agency’s decision to promulgate the final rule based on the factual record before it. As Judge Tamm noted in his original concurrence in Vermont Yankee I, that type of rule “encompass[es] both factual determinations and policy choices.”53 Although the line separating fact and policy has never been hard and fast,54 we have often employed the distinction to calibrate the intensity with which we scrutinize an agency’s contemporaneous explanation of its decisions.55 Judge Tamm explained in Vermont Yankee I how the distinction has been commonly employed:
This court has also distinguished between factual determinations and policy choices more peculiarly within the expertise of *75the administrative agency. In the former case, we commonly have insisted upon sufficient attention to the facts to enable the reviewing court to ascertain the underlying rationality of the resultant regulations. ... In contrast, “[w]here . . . the regulations turn on' choices of policy, on an assessment of risks, or on predictions dealing with matters on the frontiers of scientific knowledge, we will demand adequate reasons and explanations, but not ‘findings’ of the sort familiar from the world of adjudication.”56
Contrary to intimations in the majority opinion57 none of the specific figures in Table S-3 were intended to reflect factual certainties.58 Rather, they were a series of estimated fuel cycle impacts based on agency predictions, educated guesses, and assumptions about the future, combined with quantitative analysis. As such, they fell squarely into the gray area between “fact” and “policy.” When such predictions are asserted as facts based on hard data, with all uncertainties and assumptions underlying them undisclosed, they deserve to be scrutinized as factual determinations. When they are asserted as qualitative risk assessments, however, validly based on agency expertise and the best available information, with underlying uncertainties and assumptions fully exposed, they deserve the greater deference given to agency policy choices.59 It is impossible to determine when predictions deserve the deference given policy choices and when they deserve the closer scrutiny given facts without assessing the purposes for which the predictions were made, the methodology by which they were developed, and the conclusions they served to support. For this information we must turn to the agency’s contemporaneous explanation, as the Supreme Court directed.
B. Contemporaneous Explanation
1. The Commission’s Purpose
A review of the record reveals that the Commission’s expressed purpose in conducting the final S-3 rulemaking had evolved considerably since it had conducted the original rulemaking.60 Its expressed pur*76pose in conducting the final rulemaking was not to evaluate or select the most effective long-term waste disposal technology,' to model precisely all of the fuel cycle impacts from a single reactor, or even to evaluate the likelihood that the nuclear waste could safely be stored in the long term.
The NRC’s statement of basis and purpose for its “final” S-3 rule made clear that creation of an S-3 table had become only one part of an ongoing and wide-ranging Commission effort to evaluate long-term nuclear waste disposal issues. The Commission began by noting that “a general update of the rule with respect to all aspects of the uranium fuel cycle” was in progress.61 It further noted its intent to conduct a supplementary rulemaking directed at producing and incorporating into the final rule an “explanatory narrative” summarizing in understandable terms the environmental significance of the table’s numerical impact values.62 Finally, the Commission noted its *77intent to conduct a parallel generic proceeding — a “waste confidence” proceeding, now in progress — to “consider the most recent evidence regarding the likelihood that nuclear waste can be safely disposed of and when that, or some other off-site storage solution, can be accomplished.”63
Accordingly, the Commission’s statement of basis and purpose referred again and again to the final rule’s “limited purpose”: to provide a table modeling some, but not all, of the fuel cycle impacts for a typical reactor:64
The table of impacts adopted as a final rule in this proceeding serves as an important first step in this consideration, relieving adjudicatory boards from the need to determine those numerical impacts of the uranium fuel cycle which have been extensively considered in generic rulemaking.65
Because individual licensing boards would incorporate the table into environmental cost-benefit analyses where costs would be measured in terms of the impact of effluent releases on human lives, not in terms of the effluent releases alone, the Commission expressly stated that “use of the table in individual licensing will not foreclose discussion of the significance’’ of the release values stated in the table, or of “other *78important aspects of the fuel cycle not addressed by the table.”66
In short, the Commission’s goal was not to produce an exhaustively comprehensive table of precisely accurate figures as to which all uncertainty had been eliminated. Indeed, had the Commission set that goal as the objective of the rulemaking it could never have produced a Table S-3 at all, since all of the figures stated there were based on some unprovable assumptions, predictions, and generalizations. The Commission candidly acknowledged that its rule-making “necessarily involve[d] a wide-ranging inquiry and a certain amount of speculation.”67 Nevertheless, the Commission chose as a matter of policy not only to acknowledge that uncertainty, but to accept it, and then to proceed in the face of that uncertainty.68
The Commission gave two reasons for creating an admittedly imperfect table of model fuel cycle impact values: its concern that NEPA cost-benefit analysis demanded a prelicensing assessment of the environmental effects of the fuel cycle, generic to all licensing actions; and its determination that such assessment lay beyond the expertise of individual licensing boards:
A study of fuel cycle impact thus involves difficult generic analysis and prediction well outside the normal scope of facility-specific subjects dealt with by a reactor licensing board. This does not mean the subject can be ignored or deferred until the fuel cycle facilities themselves come up for licensing. It does mean that in reactor licensing fuel cycle impacts should be treated where possible by generic rulemaking rather than case-by-case adjudication.69
Aware that individual licensing boards would rely heavily on the table in future adjudications, however, the Commission deliberately selected its values so that, taken as a whole, the table would reflect “substantial conservatism.”70 So long as the table as a whole was conservative, the Commission assumed, it would adequately serve as a starting point in individual environmental cost-benefit analyses, even if not perfectly detailed and precise.
2. The Commission’s Methodology
To derive a conservative table of useable estimates amidst uncertainty about future technology, the NRC staff had to select some methodology. The staff chose to analyze intensively the most credible long-term waste disposal method then known — burial of the wastes in a bedded-salt geologic repository several hundred meters below ground — then “estimate[d] its impact conservatively, based on the best available information and analysis.”71 Because such a repository had not yet been built, the staff had to make a number of preliminary assumptions even to begin analysis. At the outset, it assumed that an appropriate salt *79deposit site would be found72 that the particular salt formation chosen would likely remain essentially undisturbed for millions of years,73 and that whatever repository built would be designed, operated, and regulated in such a way as to minimize the possibility of long-term failure.74
The staff then analyzed the possibility of a repository failure resulting from two types of events: (1) commonplace occurrences, such as long-term corrosion of waste cannisters and subsequent leakage into the groundwater;75 and (2) unusual occurrences, such as violent disruption of the repository by sabotage or natural catastrophe.76 Having thereby established a range of probabilities for potential waste release based on a variety of repository failure scenarios, the staff then used two approaches to assess the consequences to human life of a waste release: a “hazard index”77 and a modeling of waste transport to the human environment.78 Throughout the staff’s analysis, working assumptions were expressly spelled out and uncertainties inherent in the models employed were candidly acknowledged.79
3. The Commission’s Conclusions
The Commission reiterated the staff’s assumptions and uncertainties in its statement of basis and purpose.80 Based on the record compiled by the staff and hearing board, the Commission then found that the
characteristics of a bedded-salt repository afford a reasonable basis for the staff’s conclusion that the repository can maintain its integrity, provided that sites meeting the selection criteria can in- fact be found and developed.81
While agreeing “that areas of uncertainty remain regarding both the likelihood of finding a site and the probability that it will perform as expected,”82 the NRC noted that more precise evaluation and selection of the “best” Waste disposal technology were properly the subject of other NRC proceedings.83
Then and only then did the Commission consider whether admitted uncertainties regarding the likelihood of successful waste disposal “can or should be reflected explicitly in the fuel cycle rule”:
On the individual reactor licensing level, where the proceedings deal with fuel cycle issues only peripherally, the Commission sees no advantage in having licensing boards repeatedly weigh for them-' selves the effect of uncertainties on the selection of fuel cycle impacts for use in cost-benefit balancing. This is a generic question properly dealt with in this rule-making as part of choosing what impact values should go into the fuel cycle rule. The Commission concludes, having noted that uncertainties exist, that for the limited purposes of the fuel cycle rule it is reasonable to base impacts on the as*80sumption which the Commission believes the probabilities favor, i.e., that bedded-salt repository sites can be found which will provide effective isolation of radioactive waste from the biosphere.84
Based on staff findings of extremely low probabilities of repository failure from either commonplace occurrences or accidental intrusions and an assumption that all volatile radioactive products would be released from a repository before final sealing, the Commission found “that taking post-sealing releases as zero does not significantly reduce the overall conservatism of the table.”85 In concluding that the model chosen by the staff to assess waste disposal impacts was reasonable and adequate for the purposes of the rule, the Commission in no way foreclosed reassessment of its conelusions in light of new evidence. Observing that the “current rate of growth of knowledge [in this area] is very large,”86 the NRC reiterated that “[a] continuing reassessment of the Commission’s views on waste disposal is part of the commitment which the Commission has made to Congress.” 87
C. The Appropriate Standard of Review
The agency action challenged here was an informal generic rulemaking. Under 5 U.S.C. § 706 of the Administrative Procedure Act, which governs judicial review of agency action, the well-established standard for our review of that action is whether the agency decision was “arbitrary, capricious,' an abuse of discretion, or otherwise not in accordance with law.”88 That is the stan*81dard which Judge Tamm expected,89 and the Supreme Court directed90 us to use in evaluating the agency’s decision. Although it is also the standard which the majority purports to use, a careful reading of their opinion reveals that they have paid it lip service only. The approach actually adopted results in a judgment-substituting analysis which not only exceeds the limits of proper judicial review but which denies the presumption of regularity normally accorded an agency’s action.
In Citizens to Preserve Overton Park v. Volpe,91 the Supreme Court described the appropriate focus of judicial review under § 706 by essentially consolidating the various overlapping standards listed in § 706(2)(A), (B), (C) and (D) under a single heading — “arbitrary, capricious, or an abuse of discretion” — under which it distinguished three necessary inquiries.92 First, the Court held, a reviewing court is “required to decide whether the [agency] acted within the scope of [its] authority. This determination naturally begins with a delineation of the scope of the [agency’s] authority and discretion.” 93 Implicit in this determination is an assessment of the range of choices delegated by Congress to the agency’s discretion and a decision whether the agency’s action can reasonably be said to fall within that range of choices.
A second directive found in Overton Park was that reviewing courts carefully determine whether the agency’s actions have followed the “necessary procedural requirements.” 94 In this case, those requirements are found in both the relatively loose notice-and-comment procedures of § 553 of the Administrative Procedure Act95 and the disclosure requirements of NEPA.96
Finally, Overton Park held that the court must engage in a limited substantive review of the agency’s actual decision. Even if an agency has made a decision Congress has empowered it to make, has accorded the parties “administrative due process” by observing all required procedures, and has supported its decision with some evidence,97 *82the court must still assure itself that the agency based its result on a sustainable rationale — i.e., that the agency demonstrated a logical connection between the evidence it adduced and the conclusion it reached.98 In the words of Overton Park, § 706
requires a finding that the actual choice made was not “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” To make this finding the court must consider whether the decision was based on a consideration of relevant factors and whether there has been a clear error of judgment. Although this inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one. The court is not empowered to substitute its judgment for that of the agency.99
In recent years this court has elevated the Overton Park notion of limited substantive review of agency decisions in critical cases into a somewhat ill-defined doctrine commonly called the “hard look” doctrine of judicial review.100 While the original concept, pioneered by the late Judge Leventhal, was that agencies should take a “hard look” at complicated technological decisions with serious environmental, health, and safety consequences,101 later decisions have read the doctrine to require courts to take a hard look to ensure that agencies have taken a hard look.102 Not surprisingly, the proper level and nature of judicial, as opposed to agency, scrutiny of informal rule-makings remains a subject of intense judicial 103 and scholarly104 debate after Vermont Yankee II. Nevertheless, whether expressed in “hard look” terms or not, the limited' substantive review endorsed by Overton Park addresses legitimate concerns. As Judge Wald recently noted in National Lime Ass’n v. EPA :105
Collectively, these concerns have sometimes been expressed as a need for “reasoned decision-making” and sometimes as a need for adequate “methodology.” However expressed, these more substantive concerns have been coupled with a *83requirement that assumptions be stated, that process be revealed, that the rejection of alternate theories or abandonment of alternate courses of action be explained and that the rationale for the ultimate decision be set forth in a manner which permits the public to exercise its statutory prerogative of comment and the courts to exercise their statutory responsibility upon review.106
D. The Standard of Review Applied to the Agency’s Decision
1. “Scope of Authority”
As this court has held time and again,107 the Atomic Energy Act’s statutory delegation of authority to the NRC is unusually broad. Judge McGowan noted in Siegel v. AEC,
[In the AEA] Congress . . . enact[ed] a regulatory scheme which is virtually unique in the degree to which broad responsibility is reposed in the administering agency, free of close prescription in its charter as to how it shall proceed in achieving the statutory objectives.108
The Vermont Yankee II Court was emphatic in its reaffirmation of the NRC’s broad authority to deal with fuel cycle issues by rulemaking.109 Necessarily encompassed within that rulemaking authority is the power to treat certain factual and policy issues generically.110 And the power to treat those issues generically itself must include the discretion to incorporate a certain amount of prediction into the final rule’s table of release values.111
The NRC has confidently — and to my mind, correctly — determined that the decision to promulgate a rule in the form of a noncomprehensive table of predicted release values — whose significance for human lives is subject to discussion in individual licensing proceedings, whose accuracy is subject to update and refinement, and whose uncertainties are candidly acknowledged — lies well within the authority granted to it by Congress. Under the Supreme Court’s directive in Power Reactor Development Co. v. International Union of Electrical, Radio, & Machine Workers,112 we are obliged to *84give that determination considerable deference.113
In accordance with Power Reactor, this court has frequently held that where agency rules, such as the one here, are based on “choices of policy, . . . assessment of risks, or . . . predictions dealing with matters on the frontiers of knowledge” the agency has the authority to promulgate such rules, so long as they are based on such facts as are available and on rational predictions about how currently unanswerable questions will someday be answered.114 Thus, an agency “may apply [its] expertise to draw conclusions from suspected, but not completely substantiated, relationships between facts, from trends among facts, from theoretical projections from imperfect data, from probative preliminary data not yet certifiable as ‘fact,’ and the like.” 115 The agency does not merely have the power to promulgate a rule based on such predictions; it has a duty to do so as long as it exercises that authority “subject to the restraints of reasonableness, and does not open the door to “‘crystal ball” inquiry.’”116
Judge Bazelon does not openly challenge the broad powers exercised generally by the NRC under the authority of the AEA, but he does not expressly recognize them, either. And it is obvious that the deference due to the Commission has not tempered the majority’s determination to find some error in the Table S-3 Rule. Here, as in Vermont Yankee I, it is a NEPA-based standard of Judge Bazelon’s manufacture, bootstrapped into the APA, which provides the means to the “desired” end. Contrary to holdings in this court recognizing the flexibility of agency duties under NEPA in the face of uncertainties,117 the majority has found in its construction of NEPA procedural and substantive standards by which to limit the NRC’s broad substantive discretion. It holds that the Commission’s action failed to meet NEPA’s mandates and that it *85was taken without proper consideration of all relevant factors. These relate more to Overton Park’s second and third inquiries than to the general question of authority.
2. “Observance of Procedure”
The Vermont Yankee II Court confirmed our duty to conduct a limited procedural review of the Commission’s activity by restating the second aspect of the Overton Park analysis: “Of course, the [circuit] court [on remand] must determine whether the agency complied with the procedures mandated by the relevant statutes.”118 It warned, however, that we “should ... not stray beyond the judicial province to explore the procedural format or to impose upon the agency its own notion of which procedures are ‘best’ or most likely to further some vague, undefined public good.”119 Judge Bazelon has effectively ignored this warning, however, and once again disguises a strict substantive review under NEPA in a fundamentally procedural rubric. While our scrutiny for compliance with NEPA’s procedural mandate may be, as Judge Bazelon asserts, a close one, it must not extend across the line separating the Commission’s procedural duty from substantive determinations and policymaking. The latter category of action is properly reviewable only under the “arbitrary and capricious” standard. Thus, although I agree with the majority opinion that NEPA’s requirements are among those providing an overall procedural standard for the Commission’s action, I disagree with Judge Bazelon’s assertions of what those requirements are and how they are to be applied in the rulemaking context.
The Supreme Court emphasized in Vermont Yankee II that while “NEPA does set forth significant goals for the Nation, ... its mandate to the agencies is essentially procedural,” 120 and that “the only procedural requirements imposed by NEPA are those stated in the plain language of the Act.” 121 The plain language of NEPA, to the extent relied upon by the majority,122 requires simply that a detailed statement of environmental impacts, effects, and irretrievable resource commitments, as well as alternatives to the proposed activity, be included in recommendations for “major Federal actions.” 123 And while this “detailed statement” is only “the outward sign that the environmental values and consequences have been considered during the planning stage of agency actions,”124 the Act does not prescribe exactly how this consideration must be made. Thus, any NEPA-based review consistent with Vermont Yankee II must be limited to a scrutiny of compliance with only the “outward signs” of environmental consideration clearly prescribed by the Act.125
Notwithstanding the Supreme Court’s clear guidelines, however, Judge Bazelon has again strayed beyond judicial province to register his dissatisfaction with the methodology underlying the Table S-3 Rule and to prescribe “procedural” mandates which, however promotive of an ideal that environmental concerns play a vital role in *86government decisionmaking, simply are not among those required by Congress under NEPA.126
The majority’s first novelty is the elevation of “uncertainties” to the level of separate environmental impacts, effects, or costs. Although this is no easy task in logic, since uncertainty is a quality inherent in all impacts as well as benefits, of which there is less than complete certitude, the equation is accomplished semantically with apparent ease. Pointing to “1) uncertainty concerning the integrity of the [projected] permanent repository . . . and 2) uncertainty over whether and when such a repository ... will be developed,”127 he concludes that “these uncertainties reflect two environmental costs of licensing a plant.” 128 From that point “costs” are treated as an equivalent to “uncertainties,” and uncertainties thus become costs which must be considered under NEPA.
Judge Bazelon’s prescription does not end there, however. Under his analysis no simple consideration of costs will do: “Although the Commission did consider these uncertainties, it did not do so in a manner that would allow licensing decisions to be affected.” 129 The majority’s bottom line is that uncertainties may be fully considered only if openly balanced against benefits, as costs now traditionally are. This, it is maintained, is the type of consideration described in Calvert Cliffs’;130 and thus, we are led to conclude, is the type of consideration required by NEPA in this case. And because the Commission “did not rule that the [uncertainties] were insignificant, [or] rule that they were outweighed by generic benefits that would also be excluded from the licensing boards’ consideration,”131 and because the promulgation of a rule calling for an assumption that post-sealing repository releases would be zero prevented licensing boards from considering uncertainties in their own balancing,132 “the Commission directly contravened NEPA’s requirement that environmental costs be considered ‘at every stage where an overall balancing of environmental and non-environmental factors is appropriate.’ ” 133
Thus, in a series of small steps, Judge Bazelon has held that uncertainties in the assumptions underlying the figures in the fuel cycle rule can be considered, consistent with NEPA, only as separate environmental costs, balanced in common EIS practice against benefits. He has severely qualified and limited the Commission’s authority to determine by an otherwise reasonable generic process that certain factors do not merit separate consideration in individual *87licensing proceedings.134 Clearly, the requirements described by Judge Bazelon are not among the procedural requirements “stated in the plain language of [NEPA]; ” 135 they are “creature[s] of judicial cloth, not legislative cloth,” 136 and are an improper basis for invalidating the Rule. Our critical review of an agency’s action, such as the S-3 Table here under scrutiny, ends with a determination that the decision-makers have satisfied the key “procedural” requirement of considering environmental consequences of and alternatives to the proposed action. Even though NEPA provides that environmental considerations must be made “to the greatest extent possible,” we may not, under the guise of close scrutiny for procedural compliance, dictate to the agencies just how this consideration is best made. As each of the four major Supreme Court NEPA cases137 has shown, judicial review of agency compliance with NEPA is to be tempered by a recognition that full compliance is measured against the agency’s structuring of its own proceedings, in which the agency is to be accorded the greatest deference.
3. “Reasoned Decisionmaking”
I agree with Judge Bazelon’s statement that “a reviewing court must set aside an agency finding as arbitrary if it determines that the finding is not based upon consideration of relevant factors or if it is based on a clear error of judgment.” I strongly disagree, however, that this rule, properly applied to this case, should yield the result reached by the majority.
Although the fact is not specifically acknowledged, it is obvious from a careful examination of Judge Bazelon’s opinion that the majority’s holding against the Table S-3 Rule is rooted simply in its dissatisfaction with a single conclusion underlying the Rule: that licensing boards effectively may ignore the uncertainties inherent in any projections related to long-term storage of nuclear wastes in favor of the probability-based zero-release figure. Judge Bazelon concedes that “uncertainties can be assessed generically, and that attendant risks can be measured generically,”138 but states categorically that “the Commission cannot find the environmental cost represented by the uncertainties to be zero unless their cost is, in fact, zero.” 139 He has obviously made from the zero-release figure included in the Rule something which it is not — a finding of fact rather than simply a practical estimate for limited purposes. A careful review of the Commission’s contemporaneous explanations convinces me that there is no basis for finding this determination by the NRC to be arbitrary, capricious, or clearly erroneous; nor is it “otherwise not in accordance with law,” as the majority suggests. The majority has simply substituted its “expertise” for that of the Commission.
The relevant question with regard to the zero-release assumption is not whether the zero figure itself is precisely accurate but *88whether the Commission’s decision to adopt it was reasoned. The first step must be to recognize that the figure was never meant to be a “factual finding,” as the majority proposes;140 it is intended only as a generic determination of the appropriate weight to be accorded in licensing proceedings to risks of radioactive emission from long-term waste storage. The Commission did not decide that releases would never occur at the projected federal repository. It only decided, in view of the staff’s assumption that fission-product gases would be released prior to sealing and its finding of extremely low probability of repository failure, that “taking post-sealing releases as zero does not significantly reduce the overall conservatism of the table.”141
As noted earlier, the majority concedes that “the Commission’s zero-release assumption is probably characterized better as a decisionmaking device, ... instructing licensing boards to assume that nuclear waste will have no impact on the environment once it is sealed in a repository, [and] allocating] to the Commission sole responsibility for considering the risk that long-lived wastes will not be disposed of so successfully.” 142 Yet rather than to afford the Commission any deference in its decision to include the zero figure “device” in the generic rule, Judge Bazelon has cut to the quick of the NRC’s expertise and judgment: “For an agency to go forward in the face of significant uncertainty and issue such a rule indicates either a failure to consider a relevant factor or a clear error in judgment, . . . [and] that is precisely what the Commission did in promulgating the Table S-3 Rule.”143 The record clearly contradicts this conclusion.
There is no question that the Commission considered relevant factors — including the uncertainties with which Judge Bazelon is so concerned — in deciding to promulgate the S-3 Rule. The record is replete with evidence that the Commission took fair account of these uncertainties. Judge Bazelon in fact observes that “[f]rom the time it proposed the original Rule ... through its promulgation of the final Rule, the Commission became increasingly candid in its acknowledgment of uncertainties underlying permanent waste disposal.” 144 We can demand no more, and the majority’s holding that relevant factors were not “properly considered” because the consideration did not take a particular form or yield a usable result145 exceeds the limits of our judicial prerogative in a substantive review of agency decisions under the APA.146
The Commission’s decision not to reflect uncertainties about long-term waste storage directly in the table was simply part of its overall decision to treat an issue generically. That decision was a quintessential policy judgment.147 The Commission chose *89to incorporate into the table only those values to which it could reasonably assign a figure. It did not attempt to quantify the uncertainties surrounding the figures themselves because they were inherently unquantifiable.148 It did not require individual licensing boards to weigh repeatedly the effects of uncertainties because it was aware that individual boards were incapable of doing a better job of analysis than the Commission had already done. It did not institute a separate generic proceeding to consider uncertainties because the Commission expected that long-term effects would be more fully analyzed and understood after the concurrent “waste confidence” proceeding had concluded. In view of all of these considerations, I do not believe the Commission abused its discretion by finding “no advantage in having licensing boards repeatedly weigh” for themselves the unquantifiable uncertainties inherent in the rule it adopted- — a table whose values mixed both fact and prediction.
Nor did the Commission fail to take a hard look at the whole problem before it. Its avowed objective was to produce a conservative, noncomprehensive table of model fuel cycle impacts; its methodology was directed toward that objective. Its working assumptions were revealed, alternative theories exposed and explained, and its ultimate rationale set forth in a logical and comprehensible manner. The Commission’s decision not to search out and eliminate all uncertainties from its table can hardly be characterized as irrational in view of the limited purpose for which that table was being created.
There was no clear error in the Commission’s conclusion that the use of a zero-release figure, rather than one which reflected a post-sealing release probability which was zero to many decimal places, would have little effect on an individual board’s decision to license any given plant.149 Even after considering the uncertainties involved in its projections of long-term waste storage techniques, the NRC concluded that the probabilities (of which uncertainty is itself an element) still favored the assumptions made and that the presence of uncertainties required no modification of the judgmental zero-release figures. This conclusion did not foreclose further investigation into or consideration of the uncertainties, and it did not preclude the licensing boards from considering uncertainties which were not separately considered in the Commission’s generic determination. It was not a conclusion which was outside the realm of reasonableness. A fortiori, I cannot find the Commission’s decision to adopt the entire table, of which the zero-release figure was but one small part, arbitrary and capricious, given the compensating conservatism of the rest of the table.150
*90Nor can I say, in light of recent precedent, that the NRC has shirked its substantive obligations under NEPA. After Strycker’s Bay Neighborhood Council, our only role “is to insure that the agency has considered environmental consequences.” 151 Here the Commission satisfied its obligation to consider those consequences in two ways: by considering whether the assumption of complete repository integrity (inherent uncertainties and all) would disrupt the overall conservatism of Table S-3 and skew subsequent environmental cost-benefit analyses; and by permitting the significance to the environment of the chosen fuel cycle impact values to be the subject of discussion at individual licensing proceedings.
I cannot avoid contrasting Judge Bazelon’s surprisingly stringent review of the Commission’s conclusions concerning the technological feasibility of the assumptions underlying the values in Table S-3 with his deferential, straightforward, and proper application of the arbitrary and capricious test to the NRC’s conclusions of economic feasibility. He concedes that standards guiding conclusions concerning economic feasibility “may be based on realistic, conservative, and reasoned forecasts,”152 yet the forecasts upon which the Commission based its conclusions of technological feasibility were no less “realistic, conservative, and reasoned.” And he is willing to look to the record as a whole for evidence that the NRC considered factors relevant to dollar cost and that it “reached a conclusion that is within the range of reasonability,”153 yet is unwilling to afford the same deference to the Commission’s consideration of the uncertainties relevant to its technological predictions.
I am puzzled only by Judge Bazelon’s decision not to apply the standard which tempered his substantive review of the NRC’s economic conclusions to the case as a whole.154 If in the economic arena, then a fortiori in the technological arena, the court must accord to the Commission the deference due in this case.
Based on this review of the authority upon which the NRC acted in this informal rulemaking, the procedures it employed, and the stated rationale for its decision, I believe we have ample ground to affirm.
IV. LIVING WITH UNCERTAINTY
Having expressed my concerns with the particulars of Judge Bazelon’s analysis, I am compelled to respond specifically to the central, and equally harmful implication of the majority’s opinion. Whether his holding is based outwardly on procedural or substantive grounds, the effect of the majority’s analysis is to state that it is error for the Commission to proceed in the face of admitted uncertainties.
In our everyday lives we make countless minor and major decisions, hopefully without stepping beyond the bounds of reason, despite unfathomable uncertainties. We drive our cars without understanding fully the potential risk of injury and death; we eat, drink, and breathe without total certainty that the substances we ingest will not cause long-term disease; we plan for our futures without assurance that that fu*91ture will not be obliterated by unexpected wars, plagues, or natural disasters. Of course we try to evaluate the risks of automobile accident, fatal disease, and possible calamity as well as we can. But when those risks are truly incalculable, and potential outcomes are truly uncertain, we necessarily face a decision: do we stop until all uncertainties are resolved, or do we proceed cautiously in the face of uncertainty, making tentative assumptions and predictions, and revising and perfecting those predictions as more information becomes available?
In the nuclear power area it is Congress which has made the decision to proceed in the face of uncertainties.155 Congress has charged the NRC not only with authority and discretion to decide how best to proceed, but also with a mandate to predict likely future developments.156 As a reviewing court we are entitled to demand that the Commission’s predictions be based on logic and the best available information; we may further demand that the Commission seek out new information as it becomes available, and that it continually modify its rules to reflect the state of existing knowledge.157 A reviewing court exceeds its authority, however, when by the stringency of its review it effectively forces an agency to employ new procedures or to rewrite its rules until it reaches what the court believes is the best or correct result.158
Judge Bazelon’s opinion today invites this court to exceed the scope of its authority in just this manner. By holding that generic uncertainties must be treated as separate environmental impacts which, unless explicitly factored into a generic rule,159 must be considered anew in each individual proceeding, Judge Bazelon is dictating to the Commission when and where, and under what conditions, it must consider a policy question Congress has empowered it to decide. Yet, after Vermont Yankee II, courts must candidly acknowledge that the choice of when, where, and under what conditions the regulatory agency makes such policy determinations lies within the agency’s discretion, especially when uncertainty is rife and technology is fast-developing.
It is not clear just how the invalidation of the S-3 Rule and the remand ordered in this case will help the NRC to make' a better decision than it has already made. Whenever toxic materials are placed in a container, a risk is created that human lives will be endangered. The magnitude of that risk, however, depends on a variety of factors, including the amount and toxicity of the materials being isolated, the duration of their isolation, and the long-term integrity of their container. A comprehensive assessment of the likelihood of injury to human life from waste repository releases would require examination and evaluation of al*92most an infinite and at this time incalculable number of variables.160
Judges Bazelon and Edwards have effectively imposed upon the NRC and its licensing boards an impossible task of predicting this court’s response to their activities. The agency cannot be sure that another aspect of its rule or procedure will not be discredited in a future appeal unless it hunts down every factor it can think of and attempts somehow to qualify and evaluate its uncertainties as “environmental costs,” an equivalence which has not been thought to exist before. It goes without saying that such an effort would require an enormous commitment of agency time and resources. The only certainty achievable by this process— and it will be achieved — is delay in the development of nuclear power.
More significantly, even if the NRC chose to devote all of its time and staff to studying the risk to human life of long-term waste storage — at best a questionable allocation of agency resources 161 — it still could not root out all uncertainties.162 Even if it managed by further research to place boundary values on those uncertainties, and thus succeeded in producing a “risk value,”163 of some sort, the NRC still would have accomplished only half of its statutory task. At some point the agency would still have to decide whether or not that level of risk and uncertainty was an acceptable one.164
*93There can be no doubt that the decision whether to stop or proceed in face of a certain level of risk and uncertainty is a policy choice clearly within the agency’s discretion. Yet the majority’s opinion today amounts to nothing more clearly than a judicial proclamation that the current level of risk-assumption is unacceptable. As then-Professor (now Judge) Breyer pointed out with respect to Judge Bazelon’s Vermont Yankee I opinion, however, this kind of risk-acceptability determination disguised as “hard look” judicial review “allow[s] courts to play too prominent a role in determining how the United States will produce energy in the future.”165 Were courts to apply Judge Bazelon’s unusually stringent standard of review to other NRC rulemakings, alternative energy sources— whose production is not so carefully regulated and whose long-term environmental effects are even less well-known — will become artificially attractive to the public utilities who must choose the most economical, reliable, and accessible of the available sources of energy generation.166 Furthermore, if judges myopically focus on the risks and uncertainties peculiar to any particular type of energy, they will likely create another, equally real and equally problematic type of uncertainty — namely, uncertainty about regulatory delay which destroys the regulated firms’ incentive to innovate.167
At oral argument in this case, counsel for the petitioners assailed the Commission for failing to reflect in its rule one particular type of risk — the risk of repository leakage resulting from random drilling. According to probabilities, counsel argued, such drilling could cause a repository leak once every 2,500 years — in retroactive perspective, once from now to before the Battle of Marathon. Given that some have projected the expected life of a long-term waste repository as 250,000 years, counsel contended that the Commission had ignored a “substantial risk.” 168
*94Lord Rothschild, a former director of the British Advisory Council for Science and Technology, has made a telling point about such “apparently authoritative utterance[s]” about “substantial risks”:
There is no such thing as a risk-free society.... But there is no point in getting into a panic about the risks of life until you have compared the risks which worry you with those that don’t — but perhaps should. Comparisons, far from being odious, are the best antidote to panic.169
Even assuming the worst — that counsel is perfectly correct170 and that a repository failure caused by random drilling would in fact cost thousands and thousands of lives171 — before we decide to panic we should compare the risks described by counsel with some of the risks we normally confront, and accept, in our everyday lives.
In effect, petitioners’ counsel urges us to postpone all nuclear power plant construction because the Commission has adopted a rule which removes from individual licensing proceedings the consideration of a risk substantially less than the risk that he would die on a given day from a volcanic eruption172 or from common influenza!173 I think it safe to say that such risks have never stopped petitioners’ counsel from making either short-term or long-term plans in his everyday life. Indeed, he would probably deny that he had even failed to “consider” those risks; instead, he would probably assert that he had considered and dismissed them, just as we daily dismiss the countless possible, yet unpredictable, risks and uncertainties which fill *95our daily lives. I find it highly anomalous that counsel so fervently argues that far smaller risks should prove not merely significant, but dispositive, in an area so multifaceted and complex as the NEC’s regulation of the licensing of a light water power reactor.
V. CONCLUSION
A close examination of the majority’s opinion today reveals that it has taken no more than a giant step sideways from an analysis rejected unanimously by the Supreme Court in Vermont Yankee II. It matters little that Judge Bazelon's latest approach has taken on a slightly different form or is presented under a different banner than his analysis in Vermont Yankee I. Its focus and effect are unchanged. Vermont Yankee III does not cure the first-round defects of Vermont Yankee I; it compounds them.
Simply put, the majority has created a new standard by which to gauge agency rulemaking against NEPA. It announces that uncertainties, even though inherent in the projections underlying environmental risk identification itself, must be considered as a separate category of environmental cost. It holds further that consideration of these uncertainties, however otherwise obvious and clearly disclosed in the rulemaking record, is insufficient if not made in a particular form or yielding a “usable” result — which in the last analysis is simply a result satisfactory to this court. Such a standard is clearly not mandated by NEPA and cannot be judicially foisted upon the Commission, however noble Judge Bazelon’s intentions.
In view of the limited purposes of the fuel cycle rule, the other proceedings concurrently being conducted by the NRC, the overall conservatism of Table S-3, the massive record before us, and the rationale clearly stated in the Commission’s statement of basis and purpose, I cannot say that the Commission acted arbitrarily and capriciously in promulgating its final rule. The Commission acted well within the scope of its authority throughout the S-3 proceeding; it faithfully observed all statutorily required procedures; and it acted reasonably both in adopting the assumption of complete repository integrity and in generically considering the uncertainties surrounding long-term fuel cycle impacts.
This is a rare case indeed. The Supreme Court has already unanimously mandated both the standard and the intensity by which we should scrutinize the very agency action being challenged. Faithful observance of that mandate demands that we now find that the agency’s rule was reasonable. The majority has not done so. The decision and the stated rationale on which it is based, taken in conjunction with the recent action of this court in halting all operation at undamaged Three Mile Island Unit 1 on the basis of another completely novel theory, signal that this court has effectively taken over control of the nuclear industry. This is not the way I read either the substantive law or a judicial commission.
On Petition for Rehearing
Before BAZELON, Senior Circuit Judge, GEORGE C. EDWARDS, Jr.*, Chief Judge, United States Court of Appeals for the Sixth Circuit, and Wilkey, Circuit Judge.ORDER
PER CURIAM.Upon consideration of the petitions for rehearing filed by respondents and by intervenors Commonwealth Edison Company, et al., and Pacific Legal Foundation, it is
ORDERED, by the Court, that the aforesaid petitions are denied.
Statement of Senior Circuit Judge Bazelon as to why he voted to deny rehearing, attached.
Circuit Judge Wilkey's position is set forth in his statement attached to the en banc order entered this date.
*96Statement of Senior Circuit Judge BAZELON as to why he voted to deny rehearing.
In voting to deny these petitions for rehearing, I have carefully considered the parties’ arguments and have concluded that no issue has been raised that has not been addressed in our opinion. I think it worthwhile, however, to respond briefly to the salient points of the petitions.
The petitioners assert, first, that we have prescribed procedures by which the NRC must consider the environmental impact of fuel-cycle activities, 'and second, that we have substituted our judgment for that of the NRC. In fact, we did neither. We held only that the NRC’s environmental consideration must conform with the mandate of NEPA, as interpreted in Calvert Cliffs’ Coordinating Committee, Inc. v. AEC, 449 F.2d 1109 (D.C.Cir.1971).
In addition, the NRC states that “[t]he panel envisaged a procedure in which waste disposal uncertainties would be quantified as costs.... ” NRC Petition at 10 (emphasis added). We said nothing about quantification; we said only that uncertain environmental costs must be considered in such a manner that they can potentially affect the ultimate decision to take an action. The Commission further argues that, under NEPA, an agency should be able to assess environmental costs in the abstract and subsequently exclude them entirely from consideration when the decision is being made whether to go forward with a project. Id. at 11. We held, and I strongly believe, that this type of consideration violates NEPA, and if permitted, would eviscerate the procedural regimen that NEPA established. As this court stated in Calvert Cliffs’, NEPA requires that “environmental issues be considered at every important stage in the' decisionmaking process concerning a particular action — at every stage where an overall balancing of environmental and nonenvironmental factors is appropriate and where alterations might be made in the proposed action to minimize environmental costs.” 449 F.2d at 1118.
On Suggestion for Rehearing En Banc
Before ROBINSON, Chief Judge, BAZELON, Senior Circuit Judge, WRIGHT,* Circuit Judge, GEORGE C. EDWARDS, Jr.,** Chief Judge, United States Court of Appeals for the Sixth Circuit, and WIL-KEY, WALD, MIKVA and GINSBURG, Circuit Judges.ORDER
PER CURIAM.The suggestions for rehearing en banc filed by respondents and by intervenors Commonwealth Edison Company, et al., Pacific Legal Foundation, and Baltimore Gas & Electric Company, et al., have been circulated to the full Court and no member of the Court has requested the taking of a vote thereon. On consideration of the foregoing, it is
ORDERED, by the Court, en banc, that the aforesaid suggestions are denied.
Circuit Judge Wilkey does not ask for en banc consideration because he believes nothing should delay consideration of this case by the Supreme Court. Circuit Judges Tamm, MacKinnon, Harry T. Edwards and Bork are recused.
. Vermont Yankee II 435 U.S. at 536 n.14, 98 S.Ct. at 1207 n.14.
. 5 U.S.C. § 551 et seq. (1976).
. Industrial Union Dep’t, AFL-CIO v. Hodgson, 499 F.2d 467, 474 n.18 (D.C.Cir.1974) (emphasis added).
. Petitioners challenge the validity of the original interim, and final versions of the rule published at 44 Fed.Reg. 45,362 (1979) and codified at 10 C.F.R. §§ 51.20(e) & 51.23(c) (1981). While Vermont Yankee I addressed the original rule and Vermont Yankee II issued after the interim rule had been promulgated, this is the first case to address the validity of the final rule.
. That section requires only that the agency provide adequate notice of the proposed rule-making, 5 U.S.C. § 553(b), “give interested persons an opportunity to participate” in the rule-making, 5 U.S.C. § 553(c), and “[a]fter consideration of the relevant matter presented . .. incorporate in the rule[] adopted a concise general statement of [its] basis and purpose.” Id.
. See, e.g., 39 Fed.Reg. 1001 (1974) (emergency core cooling systems); 40 Fed.Reg. 19,439 (1975) (radiation emissions must be “as low as practicable”); 39 Fed.Reg. 43,101 (1974) (environmental impact of using mixed oxide fuels).
. The rule is reproduced in both Judge Bazelon’s and Judge Edwards’ opinions above.
. • Section 102(2)(C) requires a “detailed statement” on the environmental impact of major federal actions. As this court recently noted, however, “[tjhe EIS ... is not an end in itself, but rather a means toward the goal of better decisionmaking.” North Slope Borough v. Andrus, 642 F.2d 589, 599-600 (D.C.Cir.1980). Thus, the decision regarding how much detail to include an EIS is primarily one for the agency itself, guided by a “rule of reason.” Id. at 600. See also Scientists' Inst. for Pub. Information, Inc. v. AEC, 481 F.2d 1079, 1091 (D.C.Cir.1973); NRDC v. Morton, 458 F.2d 827, 833 (D.C.Cir.1972). All NEPA demands of an agency is that it give “reasonable consideration to all significant impacts.” North Slope Borough v. Andrus, 642 F.2d 589, 599 (D.C.Cir.1980).
Although NEPA’s own language does not require it, since this court’s decision in Calvert Cliffs’ Coordinating Comm. v. AEC, 449 F.2d 1109 (D.C.Cir.1971), NRC licensings have included a cost benefit analysis that “considers and balances the environmental effects of the facility and the alternatives available for reducing or avoiding adverse environmental effects, as well as the environmental, economic, technical and other benefits of the facility.” 10 C.F.R. § 51.20(b) (1980).
. See generally Note, The Use of Generic Rulemaking to Resolve Environmental Issues in Nuclear Power Plant Licensing, 61 Va.L.Rev. 869, 881-82 (1975).
. “Local” issues, unlike generic issues, are particularized and unique issues specific to the construction and operation of a given nuclear power plant and hence, are properly determined in individualized license adjudications. Following the recommendations of the Administrative Conference of the United States and numerous commentators, however, the NRC began the practice of generic rulemakings because the “practice of resolving generic issues in individual, adjudicatory-style hearings [had] proved a procedural failure when judged by the criteria of efficiency, acceptability, and accuracy.” Note, Judicial Review of Generic Rule-making: The Experience of the Nuclear Regulatory Commission, 65 Geo.L.J. 1295, 1301 (1977). See also Recommendations of the Administrative Conference of the United States, 1 C.F.R. § 305.73-6 (1977).
. Vermont Yankee I maj. op. at 646.
. Id. at 655.
. Id. at 641.
. The majority’s two rationales for remand were intertwined to the point of being merged. Compare the majority’s statement of the issue to be decided:
*70we are called upon to decide whether the procedures provided by the agency were sufficient to ventilate the issues,
Id. at 643 (emphasis added) with the panel’s final mandate to the Commission:
Whatever techniques ■ the Commission adopts, before it promulgates a rule limiting further consideration of waste disposal and reprocessing issues, it must in one way or another generate a record in which the factual issues are fully developed.
Id. at 654 (emphasis added).
. Vermont Yankee I concurrence at 658 (emphasis added).
. Id. at 659 (emphasis in original).
. Id. at 661 (emphasis added).
. Justices Blackmun and Powell took no part in the consideration or decision of the case. See Vermont Yankee II 435 U.S. at 521, 98 S.Ct. at 1201.
. Id. at 548, 98 S.Ct. at 1214 (emphasis added). While the Court found it “not entirely free from doubt” whether the majority had struck down the rule because of inadequate procedures or because of an inadequate record, id. at 541, 98 S.Ct. at 1210, it eventually concluded that Judge Bazelon’s majority opinion had “struck down the rule because of the perceived inadequacies of the procedures employed in the rulemaking proceedings. ... [W]e feel compelled to address the opinion on its own terms, and we conclude that it was wrong.” Id. at 541-42, 98 S.Ct. at 1210.
.Id. at 549, 98 S.Ct. at 1214 (emphasis added). In two footnotes to its opinion, the Court elaborated on its general directive. See id. at 535-37 n.14, 98 S.Ct. at 1207-08 n.14:
Upon remand, the majority of the panel of the Court of Appeals is entirely free to agree or disagree with Judge Tamm’s conclusion that the rule pertaining to the back end of the fuel cycle under which petitioner Vermont Yankee's license was considered is arbitrary and capricious within the meaning of§ 10(e) of the Administrative Procedure Act, 5 U.S.C. § 706 (1976 ed.), even though it may not hold, as it did in its previous opinion, that the rule is invalid because of the inadequacy of the agency procedures. Should it hold the rule invalid, it appears in all probability that the Commission will proceed to promulgate a rule resulting from rulemaking proceedings currently in progress. ... In all likelihood the Commission would then be required, under the compulsion of the court’s order, to examine Vermont Yankee’s license under that new rule.
If, on the other hand, a majority of the Court of Appeals should decide that it was unwilling to hold the rule in question arbitrary and capricious merely on the basis of § 10(e) of the Administrative Procedure Act, Vermont Yankee would not necessarily be required to have its license reevaluated,
(emphasis added). See also id. 435 U.S. at 549 n.21, 98 S.Ct. at 1214 n.21.
. The Court observed that “[i]n the Court of Appeals no one questioned the Commission’s authority to deal with fuel cycle issues by informal rulemaking as opposed to adjudication. Neither does anyone seriously question before this Court the Commission’s authority in this respect.” Id. at 535 n.13, 98 S.Ct. at 1207 n.13 (citations omitted). See also infra p. 533.
. Vermont Yankee II at 547, 98 S.Ct. at 1213.
. Id. (emphasis added).
. Section 101(a) of NEPA declares in broad language that it is the “continuing policy” of the federal government “to use all practicable means and measures” to achieve a variety of environmental goals. 42 U.S.C. § 4331(a) (1976). Section 101(b) then declares that the Government had the “continuing responsibility ... to use all practicable means, consistent with other essential considerations of national policy” to achieve those goals. 42 U.S.C. § 4331(b) (1976).
*72The specific procedural directives found in section 102, including the EIS requirement, see note 17 supra, are “action-forcing” procedures designed to help implement the broad goals of •section 101. See Kleppe v. Sierra Club, 427 U.S. 390, 409, 96 S.Ct. 2718, 2730, 49 L.Ed.2d 576 (1976).
. 449 F.2d 1109 (D.C.Cir.1971).
. Id. at 1112-13 & n.5. See also EDF v. Corps of Eng’rs, 470 F.2d 289, 298 (8th Cir. 1972), cert. denied, 412 U.S. 931, 93 S.Ct. 2749, 37 L.Ed.2d 160 (1973).
. Vermont Yankee II435 U.S. at 558, 98 S.Ct. at 1219 (emphasis added). The opinion continued by clarifying that the role of NEPA
is to insure a fully informed and well-considered decision, not necessarily a decision the judges of the Court of Appeals or of this Court would have reached had they been members of the decisionmaking unit of the agency. Administrative decisions should be set aside in this context, as in every other, only for substantial procedural or substantive reasons as mandated by statute, ... not simply because the court is unhappy with the result reached. And a single alleged oversight on a peripheral issue, urged by parties who never fully cooperated or indeed raised the issue below, must not be made the basis for overturning a decision properly made after an otherwise exhaustive proceeding.
Id.
. 444 U.S. 223, 100 S.Ct. 497, 62 L.Ed.2d 433 (1980) (per curiam).
. Id. at 227-28, 100 S.Ct. at 499-500 (citing Kleppe v. Sierra Club, 427 U.S. 390, 410 n.21, 96 S.Ct. 2718, 2730 n.21, 49 L.Ed.2d 576 (1976) (emphasis added).
. Vermont Yankee II435 U.S. at 548, 98 S.Ct. at 1214 (citing Vermont Yankee I maj. op. at 653) (emphasis added).
. Id. (citations omitted) (emphasis added).
. No. 76-528, reversing Aeschliman v. NRC, 547 F.2d 622 (D.C.Cir.1976).
. Vermont Yankee II435 U.S. at 555, 98 S.Ct. at 1217. In Consumers Power, opponents of a nuclear power plant had challenged an NRC construction permit on the grounds that the accompanying EIS had failed to consider certain energy conservation alternatives to plant construction, in violation of Council on Environmental Quality (CEQ) regulations. The Commission rejected their challenge for failure to meet a “threshold test” — that the proposed alternatives were as yet “reasonably available.” Until the challengers had crossed that threshold, the agency determined, it was not required to consider their proposed alternatives. Id. at 533, 98 S.Ct. at 1206.
This court, again speaking through Judge Bazelon, had reversed, holding that the NRC’s “threshold test” was both arbitrary and capricious and inconsistent with NEPA, Aeschliman v. NRC, 547 F.2d 622, 629 (D.C.Cir.1976). Once the challenger had drawn the NRC’s attention to an alternative, Judge Bazelon held, the NRC had a duty to investigate that alternative and explain why it did not deserve further consideration, id. at 628. Judge Bazelon further found the NRC’s Advisory Committee on Reactor Safeguards (ACRS) report evaluating the proposed plant’s safety insufficiently detailed to apprise the public of potential health hazards. Id. at 630-32. The Vermont Yankee II Court reversed both rulings. Vermont Yankee II 435 U.S. at 555, 558, 98 S.Ct. at 1217, 1219.
. Vermont Yankee II at 551, 98 S.Ct. at 1215 (citing 42 U.S.C. § 4332(C)).
. Id. at 553, 98 S.Ct. at 1216.
. Citing Judge Leventhal’s discussion in Portland Cement Ass’n v. Ruckelshaus, 486 F.2d 375, 394 (D.C.Cir.1973), cert. denied, 417 U.S. 921, 94 S.Ct. 2628, 41 L.Ed.2d 226 (1974), Justice Rehnquist noted that when challengers “are requesting the agency to embark upon an exploration of uncharted territory . . . ‘comments must be significant enough to step over a threshold requirement of materiality before any lack of agency response or consideration becomes of concern.’ ” Vermont Yankee II435 U.S. at 553, 98 S.Ct. at 1216.
. To make an impact statement something more than an exercise in frivolous boilerplate the concept of alternatives must be bounded by some notion of feasibility... . Common sense also teaches us that the “detailed statement of alternatives” [required by NEPA] cannot be found wanting simply because the agency failed to include every alternative device and thought conceivable to the mind of man. Time and resources are simply too limited to hold that an impact statement fails because the agency has failed to ferret out every possible alternative. ...
Id. 435 U.S. at 551, 98 S.Ct. at 1215 (citing with approval NRDC v. Morton, 458 F.2d 827, 837-38 (D.C.Cir.1972)).
. [T]he concept of “alternatives” is an evolving one, requiring the agency to explore more or fewer alternatives as they become better known and understood. This was well understood by the Commission, which, unlike the Court of Appeals, recognized that the Licensing Board’s decision had to be judged by the information then available to it. And judged in that light we have little doubt the Board’s actions were well within the proper bounds of its statutory authority.
Id. 435 U.S. at 552-53, 98 S.Ct. at 1216.
. See, e.g., North Slope Borough v. Andrus, 642 F.2d 589, 605-06 (D.C.Cir.1980) (discussing degree of detail necessary to EIS when considerable uncertainties exist):
*74The [agency] plainly cannot be expected or required to wait until the totality of environmental effects is known.
[This case] presents a record of facts and doubts that have not yet fully matured.... Uncertainty over remote hazards can be rectified as more information is collected [necessary to] ... instruct the [agency] in [its] developing view of costs and benefits. It is more logical and efficient to ask certain questions when the truth of their premises is unveiled.
See also Scientists’ Inst. for Pub. Information v. AEC, 481 F.2d 1079, 1091-92 (D.C.Cir.1973) (NEPA requirement that agency describe anticipated environmental effects of proposed action is subject to rule of reason); NRDC v. Morton, 458 F.2d 827, 837-38 (D.C.Cir.1972) (NEPA does not require agency to discuss in detail “remote and speculative” alternatives whose environmental effects “cannot be readily ascertained.”)
. 481 F.2d 1079 (D.C.Cir.1973).
. Id. at 1091-92.
. Vermont Yankee II 435 U.S. at 557-58, 98 S.Ct. at 1218-19.
. See supra p. 521.
. Vermont Yankee I concurrence at 661 n.ll.
. Compare Ethyl Corp. v. EPA, 541 F.2d 1 (D.C.Cir.) (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976), with id. at 70 (dissenting opinion) (debating whether dichotomy exists between facts and risks).
. See e.g., Amoco Oil Co. v. EPA, 501 F.2d 722 (D.C.Cir.1974); Industrial Union Dep’t v. Hodgson, 499 F.2d 467 (D.C.Cir.1974).
. Vermont Yankee I concurrence at 661 n.10 (quoting Amoco Oil Co. v. EPA, 501 F.2d at 741) (emphasis added). Drawing some initial distinctions between the factual and policy determinations made by the NRC in the original rulemaking, Judge Tamm noted that the agency’s decisions to treat the waste storage issue generically, to license reactors, and to postpone until later the question of which method of waste storage to adopt were policy or risk assessment judgments reviewable only for “clear abuse of discretion.” Id.
. See Bazelon op. at 478-481.
. The majority opinion in effect concedes this, noting that a “factual finding” interpretation of these figures, including the “zero-release” figures, is the less likely of the two possibilities— the other being that the figures were intended only as decisionmaking devices. Id. at 481.
. Looking to the future, and commanded by Congress to make policy, a rule-making agency necessarily deals less with “evidentiary” disputes than with normative conflicts, projections from imperfect data, experiments and simulations, educated predictions, differing assessments of possible risks, and the like. The process is quasi -legislative in character.
Amoco Oil Co. v. EPA, 501 F.2d 722, 735 (D.C.Cir.1974). Accord, Ethyl Corp. v. EPA, 541 F.2d 1 (D.C.Cir.) (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976) (holding that agency may promulgate rules based on qualitative appraisal of risk when scientific knowledge provides no answer to crucial issues of fact).
.The original S-3 rulemaking record on waste management and reprocessing consisted solely of a twenty-page statement by the Director of the AEC’s Division of Waste Management professing confidence that long-term waste disposal would someday be achieved. See Vermont Yankee I maj. op. at 647-51. Although the Commission had concluded almost without reflection that waste disposal would contribute insignificantly to the overall environmental effects of the fuel cycle, it permitted no further discussion of those effects in individual proceedings beyond use of Table S-3 itself.
In the wake of Vermont Yankee I however, the Commission reopened the rulemaking proceeding to supplement the record on waste reprocessing and management. 41 Fed.Reg. 34,707 (1976). The Commission Staff produced an extensive report entitled Environmental Survey of the Reprocessing and Waste Management Portions of the LWR Fuel Cycle (NU-REG-0116 (1976)) [hereinafter NUREG-0116]. That Survey discussed gaps in existing infor*76mation about waste management assessment, made educated, conservative estimates of release levels,' and included discussions of those waste management technologies currently available. 42 Fed.Reg. 13,805 (1977). The Commission also acknowledged for the first time that uncertainties existed with respect to risks from long-term repository failure. 41 Fed.Reg. 45,849-45,850 (1976). The Commission promised to consider new data in its analysis as it became available. Id.
At the same time, the Energy Research and Development Administration (ERDA) began preparation of a generic environmental impact statement on high-level waste management, a study of repository sites, and a study of interim waste storage plans, including waste solidification. In March 1978 President Carter established an Interagency Nuclear Waste Management Task Force to coordinate the activities of the various agencies conducting long-term waste research, with the ultimate goal of establishing a comprehensive federal program for long-term waste management. 43 Fed.Reg. 53,262 (1978). After hearings, that Task Force issued a report concluding that “there is no scientific or technical reason that would prevent selection of a suitable site for a waste depository.” Nuclear Reg.Rep. (CCH) No. 190, at 4 (16 Mar. 1979). See generally U. S. Dep’t of Energy, Report TID-29442, Report to the President by the Interagency Review Group on Nuclear Waste Management (1979).
Congress has also stayed abreast of the long-term waste storage problem since Vermont Yankee I. For a partial listing of Congressional hearings on spent fuel storage and nuclear waste disposal, see Hearings on Pub. Works and Energy Research Appropriations, FY 1978, Senate Appropriations Comm., Part 5, 95th Cong., 1st Sess. 397, 518, 548 (1977); Hearings on FY 1979 Budget Review Before the Senate Comm. on Environment and Pub. Works, 95th Cong., 1st Sess. 473-482, 505, 523 (1978); Hearings on Energy and Water Dev. Appropriations, Senate Appropriations Comm., 1980 (H.R. 4388) — Part 3, 96th Cong., 1st Sess. 402, 517, 647 (1979); Hearings on Nuclear Waste Disposal Before the Senate Subcomm. on Nuclear Regulation, 96th Cong., 2nd Sess. 20, 60-234 (1980); Hearings on H.R. 6390 Before the House Subcomm. on Energy and the Environment, 96th Cong., 2d Sess., 18 March 1980 (testimony of Chairman Ahearne); Oversight Hearings on Nuclear Waste Facility Siting Before the House Subcomm. on Energy and the Environment, 96th Cong., 1st Sess. 3, 71 (1979); Hearings on Nuclear Waste Facility Siting Before the House Subcomm. on Energy and Power, 96th Cong., 1st Sess. 3, 71-88 (1979); Hearings on H.R. 7418 Before the House Subcomm. on Energy Research and Production, 96th Cong., 1st Sess., 29 May 1980 (testimony of John G. Davis, Deputy Director of Nuclear Material Safety and Safeguards, NRC).
Finally, numerous states and localities have begun to regulate in the area of long-term nuclear waste management. See generally Note, Nuclear Waste Management: What the States Can Do, 1 Va.J.Nat.Resources Law 103 (1980) (describing state efforts). Before leaving office, President Carter took steps toward establishing a comprehensive Radioactive Waste Management Program in which state and local governments would play a key role in developing and implementing short- and long-term management programs. See Exec. Order No. 12,192 (12 Feb. 1980); see also 16 Weekly Comp, of Pres. Doc. 296-301 (12 Feb. 1980).
In sum, since the original S-3 rule was first challenged in this court, the problem of long-term waste disposal — a problem little understood or studied in the early 1970’s- — has become a major concern of several federal agencies, Congress, the President, and the states, all of whom are bent on finding a long-range solution to the permanent disposal problem.
. Final Rule Statement of Basis and Purpose at 45,362.
. Id. The Commission announced that the narrative would seek to explain both the environmental significance of the release values in the table as well as “important fuel cycle issues now outside the scope of the Table, including socioeconomic and cumulative impacts, where these are appropriate for generic treatment.” Id. On 4 March 1981 the Commission submitted a version of the narrative for public *77comment as a proposed amendment to the final fuel cycle rule. See 46 Fed.Reg. 15,154 (1981). See also note 66 infra.
.Final Rule Statement of Basis and Purpose at 45,363. Since Vermont Yankee II, the courts as well as the agencies had been active in the area of long-range waste disposal. Cf. note 61 supra. In mid-1978, the Second Circuit held in NRDC v. NRC, 582 F.2d 166 (2d Cir.1978) that the Atomic Energy Act did not require the NRC either to condition its licenses on the availability of a method for permanent disposal of high-level radioactive wastes or to conduct a rule-making to determine whether permanent waste disposal would be possible without undue risk to public health and safety. The Second Circuit concluded that the congressional intent underlying the Act would be satisfied so long as the Commission found “reasonable assurance” that a safe method for permanent waste disposal would be available when needed. Id at 171-75.
Shortly thereafter, in Minnesota v. NRC, 602 F.2d 412 (D.C.Cir.1979), this court upheld an NRC decision allowing two plants to expand their on-site storage of high-level wastes, in the process unequivocally reaffirming the Commission’s power to “consider the complex issue of nuclear waste disposal in a ‘generic’ proceeding.” Id. at 416-17. This court remanded for further consideration, however, the question “whether there is reasonable assurance that an off-site storage solution will be available by the years 2007-09” (the expiration dates of the plant licenses being challenged in that case).
In response to that decision, the NRC has been conducting a generic “waste confidence” proceeding, to “assess generically the degree of assurance now available that radioactive waste can be safely disposed of, to determine when such disposal or off-site storage will be available, and to determine whether radioactive wastes can be safely stored on-site” until off-site storage is available. 44 Fed.Reg. 61,372 (1979). While regarding the “waste confidence” issue as separate and different in scope and purpose from either the S-3 rule-making, the general S-3 update, or the S-3 narrative rule-making, the NRC intends to use the waste confidence proceeding to review and update the S-3 proceeding’s conclusions regarding waste disposal. 44 Fed.Reg. 45,363.
In that proceeding, the Department of Energy has filed a lengthy initial statement reviewing the state of the art on permanent waste disposal, estimating that a geological waste repository can be achieved around the year 2000. See DOE — NE—0007, “Statement of Position of the USDOE, In the Matter of Proposed Rulemaking on the Storage and Disposal of Nuclear Waste (Waste Confidence Rulemaking)” (15 Apr. 1980). The positions of the more than fifty participants in the proceeding have now been summarized in Report of the NRC Working Group on the Proposed Rulemaking on the Storage and Disposal of Nuclear Wastes (29 Jan. 1981). Comments on that study, which identifies information gaps and issues still in controversy, are currently being solicited from the rulemaking participants.
. See Final Rule Statement of Basis and Purpose at 45,363:
The rule need not be comprehensive in scope to be a useful and valid exercise of rulemaking authority. A récord is not yet available to support a comprehensive rule dealing with all generic aspects of fuel cycle impacts relevant to reactor licensing, but the Commission is free to adopt a narrower rule that for the present leaves some of these matters for consideration in individual proceedings.
(emphasis added).
. Id. (emphasis added).
.Id. (emphasis added). In its proposed explanatory narrative, the Commission now expresses the conclusion “that the fuel cycle impacts addressed by Table S-3 cannot significantly affect the cost-benefit balance for a light water reactor.” 46 Fed.Reg. 15,154, 15,155 (1981) (emphasis added). Based on that conclusion, the Commission has proposed that, once the narrative is formally adopted as part of the final rule, “no further consideration of fuel cycle impacts addressed by Table S-3 and the explanatory narrative will be required or allowed in individual reactor licensing proceedings.” Id. If supported by the supplementary rulemaking record, the Commission’s finding of “no significant effect” would relieve the NRC of any responsibility under NEPA to consider fuel cycle impacts in future licensing adjudication. For the purposes of our review today, however, I make reference only to the contemporaneous explanation offered by the Commission in support of its table, namely, the Statement of Basis and Purpose issued in 1979.
. Final Rule Statement of Basis and Purpose at 45,363.
. “A reasonable degree of uncertainty is unavoidable and is acceptable, given that basic decisions have not yet been made regarding reprocessing and the technology of waste disposal.” Id. (emphasis added).
. Id. (emphasis added).
. Id. at 45,364.
. Id. at 45,363.
. Id. at 45,368 & n.22 (noting that salt deposits have been found in 24 of the 50 states). See also NUREG-0116 at 4-73 & Ref. 53 (showing locations of significant rock-salt deposits in continental United States).
. Final Rule Statement of Basis and Purpose at 45,368 n.21 (noting that geologic formation of existing salt beds have remained undisturbed for many millions of years already). See also NUREG-0116 at 2-11.
. See NUREG-0116 at 4-77 to 4-79 (showing probable layout of bedded salt repository and describing how provisional repository would be converted to permanent repository).
. Id. at 4-86 to 4-88.
. Id.
. Id. at 4-88 to 4-89.
. Id. at 4-89 to 4-94.
. See, e.g., id. at 4-94 (listing remaining areas of uncertainty).
. See Final Rule Statement of Basis and Purpose at 45,367-69.
. Id. at 45,368.
. Id.
. Id. at 45,368-69. The Commission noted that the Department of Energy would make the initial judgment regarding which waste disposal technology to select, and that any NRC judgments regarding the likelihood of successful long-term waste disposal would be made in the waste confidence proceeding. See note 64 supra.
. Id. at 45,369 (emphasis added). The Commission’s statement that it sees “no advantage in having licensing boards repeatedly weigh for themselves the effect of uncertainties on the selection of fuel cycle impacts” is clearly consistent with the rule’s statement that licensing boards “shall not be required’’ to consider generic uncertainties. No fair reading of the Commission’s stated intent at this time supports Judge Bazelon’s conclusion that “the Commission’s final Rule, like its two predecessors, does not permit licensing boards to consider either the risk that permanent waste management facilities will not be developed, or the risk that they will fail to perform as intended if they are developed.” Bazelon op. at 475 (emphasis added).
. Final Rule Statement of Basis and Purpose at 45,369. The staff had evaluated two types of unusual occurrences which might cause accidental breach of repository containment: natural rapid events uninfluenced by humans, such as meteor strikes, and unexpected human intervention, such as sabotage or random drilling. The first type of unusual occurrence, the staff found, would likely occur only once every 50 trillion years. As for the second type of unusual occurrence, the staff found evidence that even a surface burst of a 50-megaton nuclear weapon was unlikely to breach a repository buried at 600-meter depth, and that random drilling would cause repository breach only once in 2500 years. See NUREG-0116 at 4-87.
With respect to the more commonplace form of repository breach — gradual erosion and infiltration of the repository by the surrounding groundwater — the staff’s major finding was that bedded salt is an enormously plastic, highly impermeable medium. Id. at 4-75. So long as the repository was located in a setting known to have little contact with circulating groundwater, the staff reckoned that hundreds of thousands of years could elapse before leached radionuclides would migrate to a well or surface discharge point, exposing humans to radiation via the food chain. Id. at 4-89 to 4-93.
. Final Rule Statement of Basis and Purpose at 45,369 n.27.
. Id. at n.26.
. 5 U.S.C. § 706, governing scope of review, provides,
To the extent necessary to decision and when presented, the reviewing court shall decide all relevant questions of law, interpret constitutional and statutory provisions, and determine the meaning or applicability of the terms of an agency action. The reviewing court shall—
(1) compel agency action unlawfully withheld or unreasonably delayed; and
(2) hold unlawful and set aside agency action, findings, and conclusions found to be—
(A) arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law;
(B) contrary to constitutional right, power, privilege, or immunity;
(C) in excess of statutory jurisdiction, authority, or limitations, or short of statutory right;
(D) without observance of procedure required by law;
(E) unsupported by substantial evidence in a case subject to sections 556 and 557 of this title or otherwise reviewed on the record of an agency hearing provided by statute; or
*81(F) unwarranted by the facts to the extent that the facts are subject to trial de novo by the reviewing court.
In making the foregoing determinations, the court shall review the whole record or those parts of it cited by a party, and due account shall be taken of the rule of prejudicial error.
. See supra p. 520.
. See supra p. 520.
. 401 U.S. 402, 91 S.Ct. 814, 28 L.Ed.2d 136 (1971).
. The separate standards under § 706(2)(B)-(D) thus have been rendered largely redundant by. a broad reading of (A)’s “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”
. 401 U.S. at 415-16, 91 S.Ct. at 823-24.
. Id. at 417, 91 S.Ct. at 824 (emphasis added).
. 5 U.S.C. § 553 (1976), discussed in note 14 supra.
. 42 U.S.C. § 4332(2)(C) (1976), discussed in note 17 supra.
. When the matter under review arises in the context of a trial-type proceeding conducted under 5 U.S.C. §§ 556-57, a reviewing court must ensure that the agency decision was supported by “substantial evidence” based on the record as a whole. See 5 U.S.C. § 706(2)(E), reprinted in note 88 supra. Some statutes also require that agency rulemakings conducted under the statute be supported by substantial evidence. See, e.g., Mobil Oil Corp. v. FPC, 483 F.2d 1238 (D.C.Cir.1973) (discussing Natural Gas Act).
Judicial review of the agency’s evidentiary record in an informal rulemaking like this one, however, must be guided by the “arbitrary [and] capricious” standard of 5 U.S.C. § 706(2)(A). There is little doubt that the NRC’s extensive evidentiary record here satisfies that test since the arbitrary and capricious standard is even less stringent than the lenient “substantial evidence” standard of review. See Wright, The Courts and the Rulemaking Process: The Limits of Judicial Review, 59 Cornell L.Rev. 375, 392 (1974):
If weightings on conflicting evidence need be only “reasonable” to pass the substantial evidence test, it follows that they can be less than reasonable and still survive ,the “arbitrary, capricious” test. If this is so, the latter standard subjects a rulemaker to only the most rudimentary command of rationality. In drawing empirical conclusions, he must give actual, good faith consideration to all relevant evidentiary factors. If he has in fact given serious attention to a factor, the weight which he assigns to it in his final judgments is of virtually no concern to the reviewing court.
. See, e.g., Communications Investment Corp. v. FCC, 641 F.2d 954, 972 (D.C.Cir.1980). In Trans-Pacific Freight v. Federal Maritime Comm’n, 650 F.2d 1235 (D.C.Cir.1980), cert. denied, 451 U.S. 984, 101 S.Ct. 2316, 68 L.Ed.2d 840 (1981), we summarized our narrow scope of review under the arbitrary and capricious standards as follows:
After satisfying ourselves that the agency has acted within its statutory authority and that this action was accompanied by the appropriate procedural protections and was supported by sufficient evidence, we must inquire whether the rationale of the agency is both discernible and defensible. In undertaking this inquiry, we may not substitute our judgment for that of the Commission, but may only assure ourselves that the agency decision was rational and based on a consideration of the relevant factors.
Id. at 1251 (footnotes omitted).
. 401 U.S. at 416, 91 S.Ct. at 823-24 (emphasis added).
. See, e.g., National Lime Ass’n v. EPA, 627 F.2d 416 (D.C.Cir.1980); Culpeper League for Envtl. Protection v. NRC, 574 F.2d 633, 634 (D.C.Cir.1978) (per curiam); Greater Boston Television Corp. v. FCC, 444 F.2d 841, 851, 859 (D.C.Cir.1970), cert. denied, 403 U.S. 923, 91 S.Ct. 2229, 29 L.Ed.2d 701 (1971); Pikes Peak Broadcasting Co. v. FCC, 422 F.2d 671, 682 (D.C.Cir.), cert. denied, 395 U.S. 979, 89 S.Ct. 2134, 23 L.Ed.2d 767 (1969).
. See, e.g., NRDC v. Morton, 458 F.2d 827, 838 (D.C.Cir.1972). See also Kleppe v. Sierra Club, 427 U.S. 390, 410 n.21, 96 S.Ct. 2718, 2730 n.21, 49 L.Ed.2d 576 (1976). See generally Leventhal, Environmental Decisionmaking and the Role of the Courts, 122 U.Pa.L.Rev. 509 (1974).
. See National Lime Ass’n v. EPA, 627 F.2d 416, 451 n.126 (D.C.Cir.1980) (describing evolution of this phenomenon).
. Id. at 451-54 (discussing cases).
. See, e.g., Byse, Scope of Judicial Review of Informal Rulemaking, 33 Ad.L.Rev. 183 (1981); DeLong, Informal Rulemaking and the Integration of Law and Policy, 65 Va.L.Rev. 257 (1979) [hereinafter DeLong]; McGowan, Refíections on Rulemaking Review, 53 Tulane L.Rev. 681 (1979); Rodgers, A Hard Look at Vermont Yankee: Environmental Law Under Close Scrutiny, 67 Geo.L.J. 699, 704 (1979); Breyer, Vermont Yankee and the Courts’ Role in'the Nuclear Energy Controversy, 91 Harv.L.Rev. 1833, 1834 (1978) [hereinafter Breyer].
. 627 F.2d 416 (D.C.Cir.1980).
. Id. at 453 (footnotes omitted).
. See, e.g., North Anna Envtl. Coalition v. NRC, 533 F.2d 655, 662 (D.C.Cir.1976); Citizens for Safe Power, Inc. v. NRC, 524 F.2d 1291 (D.C.Cir.1975); Union of Concerned Scientists v. AEC, 499 F.2d 1069, 1079 (D.C.Cir.1974); Siegel v. AEC, 400 F.2d 778, 783 (D.C.Cir.1968).
. 400 F.2d 778, 783 (D.C.Cir.1968).
. See supra p. 521 & note 30.
. Judge Bazelon expressly concedes this point, see Bazelon op. at 472, which Judge Tamm also specified in his Vermont Yankee I concurrence:
[T]he decision to treat the waste storage issue through generic rulemaking because it is common to all licensing decisions is clearly a policy determination within the agency’s special expertise which we should review only for clear abuse of discretion.
Vermont Yankee I concurrence at 661 n.ll.
. Judge Leventhal stated this emphatically in Minnesota v. NRC, 602 F.2d 412 (D.C.Cir.1979), discussed in note 63 supra :
No one disputes that solutions to the commercial waste dilemma are not currently available. The critical issue is the likelihood (or probability) that solutions, either ultimate or interim, will be reached in time.... We agree with the Commission’s position that it could properly consider the complex issue of nuclear waste disposal in a “generic” proceeding such as rulemaking, and then apply its determinations in subsequent adjudicatory proceedings. Where factual issues do not involve particularized situations, an agency may proceed by a comprehensive resolution of the questions rather than relitigating the question in each proceeding in which it is raised. Petitioners hypothesize the need for individualized determinations, but we think it clear that the central issue posed by petitioners — the feasibility of interim or ultimate nuclear waste disposal solutions — is one essentially common to all nuclear facilities.
... We do not dictate the procedures of the “generic” proceeding. The breadth of the questions involved and the fact that the ultimate determination can never rise above a prediction suggest that the determination [of the probability of finding a feasible solution] may be a kind of legislative judgment for which rulemaking would suffice.
Id. at 416-17 (citations omitted) (emphasis added).
.367 U.S. 396, 81 S.Ct. 1529, 6 L.Ed.2d 924 (1961).
. In Power Reactor, the Commission interpreted its own licensing provisions to allow nuclear power plant construction permits to be granted based on a less complete finding of plant safety than would have been required to grant plant operating licenses. The Supreme Court deferred to the Commission’s interpretation of its own statute, noting that
nuclear reactors are fast-developing and fast-changing. What is up to date now may not, probably will not, be as acceptable tomorrow. Problems which seem insuperable now may be solved tomorrow, perhaps in the very process of construction itself. We see no reason why we should not accord the Commission’s interpretation of its own regulation and governing statute that respect which is customarily given to a practical administrative construction of a disputed provision. Particularly is this respect due when the administrative practice at stake “involves a contemporaneous construction of a statute by the men charged with the responsibility of setting its machinery in motion, of making the parts work efficiently and smoothly while they are yet untried and new.”
Id. at 408, 81 S.Ct. at 1535 (citations omitted).
. Amoco Oil Co. v. EPA, 501 F.2d 722, 741 (D.C.Cir.1974). Accord, Industrial Union Dep't v. Hodgson, 499 F.2d 467, 474-75 & n. 18 (D.C.Cir.1974).
Judge Mikva has recently pointed out that [t]he thoroughness and persuasiveness of the explanation we can expect from the agency will, of course, vary with the nature of the prediction undertaken. “Where existing methodology or research in a new area of regulation is deficient, the agency necessarily enjoys broad discretion to attempt to formulate a solution to the best of its ability on the basis of available information.” Industrial Union Dep’t v. Hodgson, 499 F.2d 467, 474 n.18 (D.C.Cir. 1974) .... At the other extreme, this court’s inquiry into agency methodology in the physical sciences has been far more exacting “where the facts pertinent to [a] standard’s feasibility are available and easily discoverable by conventional technical means.” National Lime Ass’n v. EPA, 627 F.2d 416, 454 (D.C.Cir.1980).
NRDC v. EPA, 655 F.2d 318, 326 (D.C.Cir.1981) (emphasis added). There can be little doubt that the agency prediction here — that a long-term solution to the waste disposal problem will be found — lies within the area where existing research is deficient and the agency necessarily enjoys broad discretion.
. Ethyl Corp. v. EPA, 541 F.2d 1, 28 (D.C.Cir.) (en banc), cert. denied, 426 U.S. 941, 96 S.Ct. 2662, 49 L.Ed.2d 394 (1976). See also EDF V. EPA, 598 F.2d 62, 83-85 (D.C.Cir.1978).
. International Harvester Co. v. Ruckelshaus, 478 F.2d 615, 629 (D.C.Cir.1973). Accord, NRDC v. EPA, 655 F.2d 318, 328-29 (D.C.Cir.1981).
. See supra pp. 523-524.
. Vermont Yankee II435 U.S. at 549 n.21, 98 S.Ct. 1214 n.21.
. Id. 435 U.S. at 549, 98 S.Ct. at 1214. That the Commission satisfied the procedural requirements for informal rulemaking under the APA was virtually dictated to us at that time: “[T]here is little doubt that the agency was in full compliance with all the applicable requirements of the Administrative Procedure Act.” Id. at 549 n.21, 98 S.Ct. at 1214 n.21.
. Vermont Yankee II at 558, 98 S.Ct. at 1219.
. Id. at 548, 98 S.Ct. at 1214.
. See Bazelon op. at 476.
. 42 U.S.C. § 4432(2)(C) (1976).
. Andrus v. Sierra Club, 442 U.S. 347, 350, 99 S.Ct. 2335, 2337, 60 L.Ed.2d 943 (1979).
. Even a close scrutiny for compliance with the procedural details is tempered by a “rule of reason” — a recognition that a court can require no more than a discussion of reasonably foreseeable impacts and reasonable alternatives. See NRDC v. Morton, 458 F.2d 827, 834 (D.C.Cir.1972).
.The error in Judge Bazelon’s approach on remand is thus the same (albeit in a slightly different form) as that identified by the Supreme Court in its review of Vermont Yankee I. This court originally held that the rulemaking procedures provided by the Commission were inadequate because they were not designed to ventilate the environmental issues to the majority’s satisfaction, or to afford the best possible consideration to environmental impacts. It therefore saw fit to dictate that agencies “develop new procedures to accomplish the innovative task of implementing NEPA through rulemaking.” Vermont Yankee I at 653. This much of the Vermont Yankee I decision was unanimously reversed by the Supreme Court, which held that a court must not use NEPA to assign procedure to a rulemaking which is both beyond that prescribed by the APA and not clearly required by NEPA. The court stated,
Absent constitutional constraints or extremely compelling circumstances the “administrative agencies ‘should be free to fashion their own rules of procedure and to pursue methods of inquiry capable of permitting them to discharge their multitudinous duties.’ ” FCC v. Schreiber, 381 U.S. [279], at 290 [85 S.Ct. 1459 at 1467, 14 L.Ed.2d 383], quoting from FCC v. Pottsville Broadcasting Co., 309 U.S. [134], at 143 [60 S.Ct. 437 at 441, 84 L.Ed. 656]. Indeed, our cases could hardly be more explicit in this regard.
Vermont Yankee II435 U.S. at 543-44, 98 S.Ct. at 1211-12.
. Bazelon op. at 483.
. Id. (emphasis added).
. Id. at 483 (emphasis added).
. Id. at 483. See also note 17 supra.
. Id. at 483.
. Id. at 483.
. Id. at 483, quoting Calvert Cliffs’.
.The majority has simply couched in procedural review a disagreement with reasoned policy decisions by the Commission. Among those, made only after reasoned consideration of uncertainties, was its determination that the effect of uncertainties on the selection of fuel cycle impacts was a generic question properly dealt with “as part of choosing what impact values should go into the fuel cycle rule.” Final Rule Statement of Basis and Purpose at 45,369. A related conclusion was that, since uncertainties had been considered as part of selecting the rule’s impact values, they need not be reflected explicitly in the fuel cycle rule. These are not rulings on procedural infirmities but are by their nature substantive holdings, the result of a review process at odds with the limits of our scrutiny under the APA. See section III.D.3. infra.
. See supra p. 535.
. See Weinberger v. Catholic Action of Hawaii/Peace Educ. Project, 454 U.S. 138, 141, 102 S.Ct. 197, 200, 70 L.Ed.2d 298 (1981).
. Vermont Yankee II; Kleppe v. Sierra Club, 427 U.S. 390 (1976); Flint Ridge Development Co. v. Scenic Rivers Ass’n, 426 U.S. 776, 96 S.Ct. 2430, 49 L.Ed.2d 205 (1976); Aberdeen & Rockfish v. SCRAP Comm., 422 U.S. 289, 95 S.Ct. 2336, 45 L.Ed.2d 191 (1975).
. Bazelon op. at 484.
. Id.
. Id. at 478-481.
. Final Rule Statement of Basis and Purposes at 45,369.
. Bazelon op. at 481.
. Id. at 485.
Id. at 481 n.118 (emphasis added).
. See id. at 484.
. As this court observed in North Slope Borough v. Andrus, 642 F.2d 589 (D.C.Cir.1980),
Given a “fully informed” and “well-considered” decision, an agency decision is entitled to judicial deference; in other words, an informed and deliberate agency decisionmaker is entitled to be arguably “wrong,” a court’s differing view of wisdom in administration notwithstanding. Obedience to NEPA is a matter of the administrative agency acquiring and digesting useful information about the environmental ramifications of major federal projects.
Id. at 599.
. As such, the decision hardly seems in violation of NEPA. As one commentator has noted:
Decisions based on extreme uncertainty are quintessential policy judgments. Although the nature and degree of risk underlying an official endorsement of continued nuclear waste generation may be highly contested issues, the correctness of the official position is not assailable in the courts on policy grounds. NEPA sets no standards for factual predicate sufficiency. If the full scope of uncertainty is exposed and duly considered, the NEPA mandate is satisfied. Dissatisfied rulemaking participants may enlist the aid of the courts if they can prove that agency officials are being disingenuous about the unknowns involved, but they must turn *89to avenues of political redress if they simply disagree that the full data base is sufficient to predict a negligible environmental impact.
Comment, Implementing the National Environmental Policy Act through Rulemaking, 126 U.Pa.L.Rev. 148, 190 (1977) (emphasis added).
. Cf. Union of Concerned Scientists v. AEC, 499 F.2d 1069, 1093 (D.C.Cir.1974) (“Considering the unpredictable nature of some of the factors [before the Commission] ... it would actually be surprising if it had attempted to incorporate the probability of their occurrence into its estimate of environmental impact.”)
. This court has long recognized that when an agency’s environmental impact statement
involves educated predictions rather than certainties, it is entirely proper, and necessary, to consider the probabilities as well as the consequences of certain occurrences in ascertaining their environmental impact. There is a point at which the probability of an occurrence may be so low as to render it almost totally unworthy of consideration.
Carolina Envtl. Study Group v. AEC, 510 F.2d 796, 799 (D.C.Cir.1975) (emphasis added).
.Indeed, two Commissioners who expressed concern about the zero release figure concurred in the final statement of basis and purpose for just the reasons I have given. Commissioner Bradford concurred in the table “with the understanding that it is to be extensively supplemented” by the explanatory narrative and the update proceedings, and concurred in the zero-release figure because it appeared “better founded than the same figure in the present interim version and because, as the Commission states, this assumption does not appear to affect the S-3 table’s overall conservatism." Final Rule Statement of Basis and Purpose at 45,372 (Separate Views of Commissioner Brad*90ford on S-3) (emphasis added). Commissioner Gilinsky, also stating his views separately, expressed similar concern about the zero figure, but noted that the forthcoming “waste confidence” proceeding was the “appropriate vehicle for a thoroughgoing evaluation of the problems involved in the government’s commitment to a waste disposal solution and the likelihood that such a program is not only feasible but is also on course.” Id. at 45,374 (Separate Views of Commissioner Gilinsky on Final Adoption of S-3 Rule).
. See supra p. 522.
. Bazelon op. at 491.
. Id. at 494.
. The Commission’s working assumption that safe waste-disposal methods are technologically feasible is no less reasoned than its assumption that economic resources will be available to meet disposal needs. There is no justification for labeling either conclusion arbitrary or capricious in this case. Both involve a mixture of factual and policy judgments, many of which the Commission is in the best position to make.
. See supra p. 524.
See also Judge Edwards’ concurring opinion, which admits that
[i]t is not the function of the judicial branch either to initiate or to halt the development of a nuclear power industry. These national policy choices are constitutionally vested in the President and the Congress of the United States, and they have been made in favor of proceeding.
Edwards’ op. at 2.
. Cf. NRDC v. EPA, 655 F.2d 318, 334-36 (D.C.Cir.1981) (discussing agency’s duty to predict).
. In Geller v. FCC, 610 F.2d 973 (D.C.Cir.1979), we held that the FCC was required to reconsider rules issued in 1972 because the factual predicate upon which the regulations had been based had changed. Similarly, in United States Steelworkers of America v. Marshall, 647 F.2d 1189 (D.C.Cir. 1980), we applied similar reasoning to the question whether an agency had discretion not to implement new standards if standards originally thought “feasible” turned out not to be so:
[Wjhere employers believe that a standard originally determined to be feasible has proved infeasible, they could petition the agency to initiate a new rulemaking to review the standard. ... [Wjhere time does demonstrate the infeasability of a standard once approved as feasible, so that the very predicate for the original rule and its statutory basis have disappeared, a court may well be able to deny the agency any discretion to refuse a new rulemaking.
Id. at 1273 (emphasis added).
. See Vermont Yankee II 435 U.S. at 546, 1213.
. See, e.g., Bazelon op. at 484.
. As one recent analysis of nuclear power decisionmaking has noted,
An appreciation of the associated hazards [of long-term nuclear waste disposal] requires an understanding of the long-term integrity of the geologic environment into which wastes are placed, the seismic integrity of rock formations, the rates of exchange between ions in ground water and in dissolved wastes, and the transport rates of water through the ground. Moreover, the waste materials have toxicities ranging from negligible to lethal in microscopic quantities, and there is controversy over the dose-mortality relationships involved. There also exist unanswered questions of rock mechanics, large-scale hydrology, and uptake of toxic metals by vegetation.
Yellin, High Technology and the Courts: Nuclear Power and the Need for Institutional Reform, 94 Harv.L.Rev. 489, 533-34 (1981) [hereinafter Yellin].
. Professor Yellin’s penetrating study of judicial review of nuclear power decisionmaking extensively surveys existing scientific studies on long-term waste disposal and concludes that that problem should rank relatively low on the list of NRC environmental concerns:
The NRC is justified in concluding that the disposal of nuclear wastes is not intrinsically a major concern. Permanent geologic disposal of wastes can create widespread contamination of ground water and soil, but at very low levels of radioactivity. On emplacing wastes hundreds of meters or more below ground, the earth itself becomes a substantial barrier to contamination of the biosphere, effective for centuries or more.... [Assuming reasonable efforts are made to choose a site isolated from ground water in a suitable medium, with waste dispersed in a geometric configuration that in itself will guard against water intrusion and ground diffusion, the dominant risks are associated with migration over millenia of heavy isotypes, principally plutonium-239, and of a few fission products, such as iodine-129 and technetium-99. While it is conceivable that these materials will cause low-level, widespread ground and water contamination, no major risk to individuals is foreseeable.
Id. at 541 n.317 (citations omitted) (emphasis added).
. The words of the Research Director of the Administrative Conference of the United States seem particularly apt here:
Anyone confronted with uncertainty naturally tends to hope that another study or consideration of an additional set of factors will resolve the issues. Information is not costless, however, and an administrator allocating a budget has to focus on the value of the information versus the costs of collection. A court has no practical incentive to weigh such costs in deciding whether an agency should have investigated more thoroughly and as a result courts might tend to demand the collection of information even when the decision at issue is not sensitive to the additional information.
DeLong, supra note 104, at 310 (emphasis added).
. Risk values are never exact numbers, but rather imprecise representations of a range of hazards considered probable. Agencies usually compute levels of risk by assuming a conservative set of assumptions, then setting boundary levels of risk which establish the range of probabilities for a particular type of event. Id. at 346-47.
. [I]n most cases [where a policymaker is stuck with a choice] the relevant datum for the decision is precisely the fact that the matter is uncertain. In other words, he must
*93confront the uncertainty rather than convert it into a fictional certainty. The crucial questions are- how to narrow the uncertainty and decide the proper course under conditions of uncertainty, given the consequences of error.
... The issue for the court is whether the agency structured its decision to highlight uncertainty problems and deal with them rationally.
Id. In this case, the Commission expressly chose to accept the “risk of uncertainty” at the generic rulemaking stage. See text and accompanying note 68 supra. The only question before us is whether that decision was an abuse of discretion. See note 147 supra.
.[l]t is reasonable to conclude that court-imposed delays in the process of nuclear plant approval are not “neutral.” They affect the decisionmaking process, tending to move the decisions of those who actually build power plants in the direction of coal, which may well prove more hazardous and environmentally destructive than nuclear plants. In these circumstances it is wrong for the courts to impose a specially stringent standard of review that requires regulatory agencies to take a specially “hard look” at all relevant factors before approving nuclear power plants. This would allow the courts to play too prominent a role in determining how the United States will produce its energy in the future.
Breyer, supra note 104, at 1840 (emphasis in original).
One commenting physicist has noted that notwithstanding that the environmental effects of wastes from coal-burning plants remain largely unquantified, the sulphur oxides from a single plant [emitted at a rate of about 10 pounds per second], are estimated to cause about 25 fatalities and 60,000 cases of respiratory disease ... [and] about $25 million in property damage annually.” Cohen, High Level Radioactive Waste, 21 Nat.Resources J. 703, 703 (1981). By comparison, “[estimates show that radioactive emissions from a nuclear plant may cause about one fatality every 50 years.” Id. at 704.
. See Breyer, supra note 104, at 1845.
. The impact of regulatory uncertainty on the incentive of public utilities to innovate has been well-chronicled. See, e.g., Cohen, Innovation and Atomic Energy: Nuclear Power Regulation, 1966-Present, 43 Law & Contemp.Probs. 67 (1979) (time, expense, and uncertainty in acquiring NRC licenses are key factors inhibiting innovation); Braeutigam, The Effect of Uncertainty in Regulatory Delay on the Rate of Innovation, ■ 43 Law & Contemp.Probs. 98 (1979).
. I must assume that this particular risk is one which Judge Bazelon would include among those reflected in his “uncertainty concerning the integrity of the permanent repository.” See Bazelon op. at 483.
. Rothschild, Coming to Grips with Risk, Wall St.J. (13 Mar. 1979) at 22, col. 4 (emphasis in original) [hereinafter Rothschild].
. Professor Yellin points out the deceptiveness of citing 250,000 years — ten times the half-life of plutonium-239 — as the appropriate-figure for measuring repository life. By citing the same figure in Vermont Yankee I, Professor Yellin suggested, this
court responded to perhaps the public’s deepest anxieties concerning nuclear energy, failing to realize that a defensible index of the hazards of radioactive waste does not follow simply from the long half-lives of some of the more toxic waste constituents, but from a complex of considerations such as the repository design, the biologic effects of remnant radioactive species, and possible modes of releasing the wastes into the biosphere.
Yellin, supra note 160, at 538-39.
I would further suggest the irrelevancy of worrying 250,000 years into the future. I, for one, find it difficult to plan that far ahead.
. In fact, a careful summary of the available scientific data suggests that this would not be the case:
Unlike major reactor accidents, .. . release of wastes from a mined repository cannot cause catastrophic events threatening public health. At worst, it can lead to widespread contamination of groundwater and soil at very low levels of radioactivity.
Id. at 534-35.
Professor Cohen notes that, even if ground water were contaminated at the projected depths of a repository,
ground water moves quite slowly, typically only inches per day, . . . [and] ordinarily travels many miles before reaching the surface from 2,000 feet underground. Hence, even if dissolved radioactive material moved with the ground water, it would take about 1,000 years to reach the surface. Additionally, processes which constantly filter the radioactive materials out of the ground water cause the material to migrate about a thousand times slower than the water itself. Thus, it would take most of the radioactive materials about a million years to reach the surface even if it were already dissolved in ground water. Moreover, most of the radioactive materials are highly insoluble under most geological conditions. If the materials were in solution when the water encountered normal conditions (chemically reducing, alkaline), they would precipitate out and form new rock material.
Finally, if radioactivity did reach surface waters it would very easily be detected. One millionth of the amounts that can be harmful are readily detected. Measures could be taken to prevent the waste from getting into drinking water or food.
Cohen, supra note 165, at 714-15.
. The eruption of Krakatoa in 1883, which killed some 36,000 humans, was not even the worst volcanic eruption to occur in the last 2500 years. See Encyclopedia Brittanica, vol. K at 498-99.
. Lord Rothschild estimates the risk of dying from influenza in a single year is 55 times as great as that of dying as a result of a nuclear power station disaster in a country where there are 100 nuclear plants. See Rothschild, note 169 supra, at 22 col. 6.