concurring in part and dissenting in part.
In my view, the district court correctly determined that the Navy abused its discretion and violated NEPA by failing to monitor new information concerning the potential environmental effects of Project ELF, and the court properly ordered the Navy to prepare a supplemental EIS. However, I agree that the district court erred when it refused to consider the countervailing equities in deciding to enjoin work on Project ELF. In light of those equities, an injunction is not warranted in this case; I therefore concur in the portion of the judgment reversing the injunction.
THE NEPA VIOLATION
In concluding that the Navy has not violated NEPA, the majority has overstepped the bounds of appellate review in its treatment of the evidence and undermined the role of NEPA when agencies undertake long-term projects in the face of uncertain environmental effects.
The majority’s treatment of the evidence in the case is both extraordinary and misleading. It is extraordinary because the majority engages in virtually de novo review of sharply disputed expert testimony. Judge Crabb’s analysis of the factual evidence on the effects of extremely low frequency radiation was thorough and thoughtful, and she is obviously much more familiar with that information than we. See Wisconsin v. Weinberger, 578 F.Supp. 1327, 1341-55, 1361-63 (W.D.Wis. 1984). Judge Crabb reviewed an extensive administrative record and heard testimony concerning controversial scientific research, and her conclusions from that evidence are precise and sound.1 Yet the ma*429jority’s discussion of the evidence in the case treats as authoritative any statements made by the Navy’s experts, Drs. Justesen and Straub. With all due respect, we are simply not in a position to reweigh that evidence and decide whose experts were more credible.2
The majority’s deprecating treatment of the plaintiffs’ evidence is misleading because it creates the appearance of disagreement where there is none and diverts attention from the central dispute. The majority and Judge Crabb and I all agree that the new scientific information alone, in the absence of other factors, is not so significant as to require a supplemental EIS. The disagreement concerns the legal relevance of the other factors Judge Crabb relied on to order the supplemental EIS.
All parties agree that the Navy has a binding duty to supplement its original environmental impact statement when there is “significant ... new information relevant to environmental concerns and bearing on the proposed action or its impacts.” 40 C.F.R. § 1502.9(c)(ii) (1983).3 The majority interprets this language to mean that the new information must present a “seriously different picture of the environmental landscape such that another hard look is necessary.” Based on its own review of the evidence, the majority concludes that the “seriously different picture” standard has not been met. Most important, the majority argues that the Navy could have violated NEPA only if it failed to issue a supplemental EIS in the face of new information meeting the majority’s new “seriously different picture” standard. Because it is clear that the new information alone does not satisfy that standard, the majority finds no violation of NEPA. The majority’s approach glides over the central problem in the case and unduly restricts NEPA in a situation where the government acts in the face of uncertain potential environmental consequences. It also conflicts with the Ninth Circuit’s decision in Warm Springs Dam Task Force v. Gribble, 621 F.2d 1017 (9th Cir.1980).
The central problem here concerns the NEPA obligations of agencies which choose to act in the face of uncertain or incorrect environmental information. In my view, Judge Crabb properly concluded that the Navy violated NEPA by failing to undertake “a thorough and comprehensive review” of the information generated since 1977 on the biological effects of electromagnetic radiation. 578 F.Supp. at 1364. The regulations and statute offer little direct guidance concerning new information and ongoing projects, and Judge Crabb correctly chose to follow a line of analysis first developed by the Ninth Circuit in Warm Springs Dam, supra. There the court said that under 40 C.F.R. § 1502.9(c), the same regulation at issue here:
When new information comes to light the agency must consider it, evaluate it, and *430make a reasoned determination whether it is of such significance as to require implementation of formal NEPA filing procedures. Reasonableness depends on such factors as the environmental significance of the new information, the probable accuracy of the information, the degree of care with which the agency considered the information and evaluated its impact, and the degree to which the agency supported its decision not to supplement with a statement of explanation or additional data.
621 F.2d at 1024. In short, the agency has a duty under NEPA to continue to monitor relevant new environmental information, and it must act reasonably to gather and evaluate that information. A decision not to undertake a supplemental EIS must be rationally based on a careful review of the new information.
Judge Crabb’s analysis of the new information here followed the Ninth Circuit’s discussion in Warm Springs Dam. Although Judge Crabb expressly declined to conclude that the new information is so significant that the Navy would have been required in the first instance to file a supplemental EIS in 1981, she concluded that the new information is “significant enough to require careful review by the Navy.” 578 F.Supp. at 1362. The research results do not prove that exposure to extremely low frequency radiation will cause adverse biological effects, but the new results do not merely reconfirm the Navy’s 1977 judgment that any environmental effects would be negligible. Id. Although the research results remain “equivocal,” the new information casts significant doubts on the Navy’s original assumption that Project ELF will not adversely affect life around it. Judge Crabb concluded that in light of the new information, “the Navy must undertake a considered review of all that is known to date and evaluate the risks and benefits before proceeding with Project ELF.” 578 F.Supp. at 1363.
Judge Crabb’s conclusions concerning the new evidence are quite similar to the Ninth Circuit’s in Warm Springs Dam. There the court held that the new information was “not so definitive as to compel initiation of the formal supplementation process,” but that it “raised sufficient environmental concerns to require the [agency] to take another hard look at the issues.” 621 F.2d at 1025. In this case, Judge Crabb did not decide that the new information mandated a supplemental EIS in 1981; she held only that the Navy was required to give the new information a careful and reasoned review. 578 F.Supp. at 1362-63.
Judge Crabb then turned to the other factors discussed in Warm Springs Dam. She examined the degree of care with which the Navy had considered the new information, and she found that the Navy had not conducted a thorough and comprehensive review of the new scientific information. 578 F.Supp. at 1364. She then noted that the Navy had never explained its decision not to prepare a supplemental EIS. On the basis of the significance of the new information, the Navy’s failure to conduct a thorough and comprehensive review of that information and the Navy’s failure to explain its decision, Judge Crabb concluded that the Navy had breached its duty under NEPA to monitor new, relevant environmental information during the life of Project ELF.
The central issues here are whether NEPA actually imposes the duty that Judge Crabb found had been violated, and, if so, whether a supplemental EIS is the appropriate remedy for that violation.
The Navy concedes that it has a continuing duty under NEPA to monitor environmental information concerning its ongoing projects. Brief for the Appellants at 48. That concession is in accord with decisions of the Ninth Circuit, see Warm Springs Dam Task Force v. Gribble, supra, 621 F.2d at 1023-26; California v. Watt, 683 F.2d 1253, 1267 (9th Cir.1982), reversed on other grounds, — U.S. —, 104 S.Ct. 656, 78 L.Ed.2d 496 (1983), as well as the First and Second Circuits, see Massachusetts v. Watt, 716 F.2d 946, 948-49 (1st Cir.1983); Sierra Club v. United States Army Corps of Engineers, 701 F.2d 1011, *4311034-37 (2d Cir.1983). See also Humane Society v. Watt, 551 F.Supp. 1310, 1322 (D.D.C.1982) (following Warm Springs Dam), affd mem. 713 F.2d 865 (D.C.Cir.1983). This continuing duty to monitor new environmental information in ongoing projects is not explicit in the language of NEPA or the CEQ regulations, but it is the implicit and necessary corollary to the express duty to supplement an EIS when significant new information arises. See 40 C.F.R. § 1502.9(c)(ii). Indeed, an agency simply cannot fulfill its duty to supplement an EIS without continuing to monitor sources of new and relevant information.
In Warm Springs Dam, the Ninth Circuit made it clear that this duty to monitor new information is not limited to information that rises to the level of “significance” requiring a formal supplement to the EIS. The court held that the agency’s decision not to supplement the EIS was unreasonable when the agency had not taken a “hard look” at a new study which called into question the agency’s prior conclusions. 621 F.2d at 1025.4 NEPA is essentially a procedural statute, and the Ninth Circuit correctly held that an agency abuses its discretion and violates NEPA when its process for deciding whether to supplement the EIS is irrational, i.e., when it does not carefully consider new and relevant information. Even if the new information ultimately turns out not to warrant a full-scale supplemental EIS, an agency still acts arbitrarily and capriciously in violation of NEPA when it ignores relevant new information or quickly sweeps it under the carpet.
There is, of course, some difficulty in discerning the precise scope of this implied duty to monitor new environmental information during an ongoing project. However, reasonable guidelines can be derived from the terms and goals of NEPA and the limited role of judicial review under NEPA. The primary purpose of NEPA is to require federal agencies to make rational decisions about the environmental consequences of their actions, and the method for achieving that purpose is to require agencies to include at all significant stages in the decisionmaking process careful and explicit consideration of those consequences. See Weinberger v. Catholic Action of Hawaii/Peace Education Project, 454 U.S. 139, 143, 102 S.Ct. 197, 201, 70 L.Ed.2d 298 (1981). The role of the courts under NEPA is limited to ensuring that agencies observe proper NEPA procedures and give due consideration to the environmental consequences of their decisions. Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, Inc., 435 U.S. 519, 558, 98 S.Ct. 1197, 1219, 55 L.Ed.2d 460 (1978).
In the case of long-term projects where the environmental effects are uncertain, the rational decisionmaking process demanded by NEPA requires that an agency continue to monitor new information which may be relevant to the uncertain environmental effects. See Southern Oregon Citizens Against Toxic Sprays, Inc. v. Clark, 720 F.2d 1475, 1480 (9th Cir.1983). If new information might significantly alter the balance of costs and benefits, an agency acts irrationally and abuses its discretion if it disregards such information. See Conservation Law Foundation v. Watt, 560 F.Supp. 561, 570-71 (D.Mass.1983), affd sub nom. Massachusetts v. Watt, 716 F.2d 946, 949-50 (1st Cir.1983). That is the situation we face-here. The 1977 EIS was based on the finding that there was no substantial evidence pointing to harmful biological effects from extremely low frequency radiation. In the meantime, a number of scientific studies of controversial validity have suggested that the radiation may indeed produce harmful biological effects. No court, obviously, can resolve that scientific dispute. But the courts are in a position to insist that the agency give the information rational and serious consideration rather than mere cursory dismissal.
*432That is precisely what Judge Crabb did here. She found that the new information was significant enough to require the Navy to give it careful consideration, and she found that the Navy had not done so. To rebut Judge Crabb’s conclusion, the Navy and the majority rely on the Environmental Review Committee, research supported by the Navy and a Navy contract with an outside consulting firm for literature reviews. Such measures are the sort of steps which might fulfill the agency’s duty to monitor new information, but upon closer scrutiny of the Navy’s actions in this case, Judge Crabb was properly unimpressed. As she discussed in her opinion, the Environmental Review Committee gave only cursory consideration to potential adverse biological effects. 578 F.Supp. at 1363. In addition, the Navy never attempted to use the research it sponsored by pulling the results together in a comprehensive review, and only one of the Navy-sponsored studies was even subjected to the ordinary scrutiny of peer review to examine the validity of the research methods and conclusions. Id. Finally, the outside consulting firm’s literature reviews were conducted only sporadically and only after the Navy had decided to go forward without supplementing the EIS. Judge Crabb therefore properly concluded that the Navy’s actions did not fulfill its duty to monitor new information and seriously consider significant new information.
NEPA does not establish — and the courts cannot prescribe — precise procedures for fulfilling this duty. The standard of review is not an exacting one here. Especially when the new information is a steady trickle of scientific studies of uncertain validity and import, it would be virtually impossible to identify a moment at which the new information becomes “significant” enough to require a supplemental EIS. An agency need only take reasonable steps, in light of the potential for significant new information, to ensure that it will consider relevant new information concerning the environmental impacts of the agency’s actions. But under this less than rigorous standard of review, the Navy’s actions still fall short, for its efforts to monitor new information were, as Judge Crabb found, merely hollow substitutes for serious consideration of potentially troubling information.
After Judge Crabb had found a NEPA violation, she was fully justified in ordering the Navy to undertake a supplemental EIS. In light of the Navy’s past failure to consider the new information seriously, it is reasonable to view with some skepticism a new but entirely in-house review of the information. Although the information here might not alone have warranted a formal supplement in the first instance, the Navy’s violation justified the remedy of a formal supplement. The supplementation process is subject to public scrutiny and criticism. That scrutiny is the best available means for ensuring that the Navy does indeed give the new information full consideration.
For these reasons I dissent from the portion of the judgment reversing the order to supplement the EIS.
THE INJUNCTIVE REMEDY
I agree with the majority that NEPA does not prohibit a court from undertaking a traditional balancing of the equities in deciding whether to issue an injunction to remedy a NEPA violation. If an injunction would not remedy the violation and promote NEPA’s goals, or if an injunction would submerge competing public values of overriding importance, NEPA does not always require that an injunction issue. The district court erred when it concluded that NEPA prohibited it from considering any competing equities.5
However, the majority overstates the case. NEPA does indeed presume that an injunction should issue in an ordinary NEPA violation case. See Massachusetts v. Watt, supra, 716 F.2d at 952; Realty *433Income Trust v. Eckerd, 564 F.2d 447, 456 (D.C.Cir.1977). When NEPA is violated, the fundamental harm is that the decision process has been tainted by the failure to consider fully the relevant environmental facts. The most effective way to remedy the harm is to freeze the situation until the agency reconsiders properly. If that “purification” of the decisionmaking process is to be at all effective, the court must ordinarily prevent the agency in the meantime from committing even more resources to the path previously chosen. As Judge Breyer pointed out in Massachusetts v. Watt, supra:
The agency as well as private parties may well have become committed to the previously chosen course of action, and new information — a new EIS — may bring about a new decision, but it is that much less likely to bring about a different one. It is far easier to influence an initial choice than to change a mind already made up.
716 F.2d at 952 (emphasis in original). Even an injunction may, as a practical matter, be inadequate when the agency is already far down the road. However, less stringent remedies will, in most cases, not be adequate to correct the NEPA violation. Courts therefore ordinarily issue injunctions in NEPA cases as a matter of course.
Nevertheless, when an agency raises powerful countervailing equities, NEPA does not prevent the district court from weighing those factors in deciding whether to issue an injunction. See Massachusetts v. Watt, supra, 716 F.2d at 951-52. Courts must recognize, though, that in doing so there is a clear risk that NEPA’s goals may be too easily diluted and subverted. NEPA protects intangible values, and competing equities may often be immediate and tangible, and too difficult to analyze critically. Only in the unusual case should competing equities prevent an injunction.
The majority has properly identified most of the factors which make this such an unusual case. First, the degree of additional commitment to the project, in terms of both time and money, during the time needed to prepare a supplemental EIS would be quite marginal here when compared with the whole project. And a substantial portion of the Navy’s additional commitments to Project ELF can, in a sense, be laid at the plaintiffs’ doorsteps because they did not file this lawsuit until nineteen months after the Navy reactivated the project. Second, the new information here is at best equivocal — it is, as a practical matter, highly unlikely to affect the Navy’s decision at this time. The information demands further study, as I have argued above, but it is too thin a reed to support an injunction closing down the project for over a year. Third, the underlying environmental threat here would not result directly from the construction of Project ELF. The only alleged environmental harms are those that might result from the eventual, long-term operation of the ELF system. An injunction here would serve only to delay the project while the Navy considered information which is at best indeterminate and equivocal concerning the possibility of environmental harm years or decades away.
The majority relies primarily, however, on the alleged harms the injunction might cause to the national defense. The national defense is without question a very important consideration and in some circumstances it could be decisive. Here, however, I do not believe it is necessary for the majority to accept the defendants’ representations on this subject at face value to reaeh the proper result on this issue. It is quite clear that NEPA does not provide a “national security” exception,6 yet the majority’s apparently uncritical acceptance of the Navy’s claims might be an invitation in some circumstances for the government to escape NEPA’s normally stiff remedies merely by asserting that an injunction might harm national security. After all, *434courts are simply not equipped to question such assertions. But the national security does not normally require such unquestioning solicitude from the courts in NEPA cases. Where national security concerns are truly paramount, Congress may, in most instances, simply exempt a project from NEPA’s requirements. It has done so in the past and will presumably do so again in the future. It is also possible that some future case would require us to weigh national defense needs unequivocally in the equitable balance, and we might then need to base our decision on the government’s representations in this respect. However, to the greatest extent possible, we in the federal courts should best leave such judgments to the political branches of government for resolution through political processes, rather than merely accept assertions which we are not in a position to challenge.
We need not entangle ourselves in speculation about the urgency of Project ELF; a far stronger basis for reversing the injunction here is that this case involves new information and an ongoing project. Although injunctions should ordinarily issue to remedy NEPA violations, special remedial problems arise with respect to new information affecting ongoing projects. The central purpose of NEPA is to ensure that agencies in fact give due consideration to environmental factors, and, as I have argued above, NEPA also imposes a duty on an agency to monitor and consider new information relevant to an ongoing project. If a project must be shut down pending preparation of a supplemental EIS, agencies will have, as a practical matter, strong incentives to subvert NEPA by merely creating a paper trail and by viewing any new information as too insignificant to require a supplemental EIS.
In cases involving new information and ongoing projects, the NEPA ideal of careful environmental planning prior to action is simply not achievable. The achievable goal is that agencies will indeed give their decisions fresh, hard looks in light of new information. The courts should not undermine NEPA by demanding more, by enjoining further work while the agency considers new information. Otherwise, the temptation to soft-pedal relevant information and to comply with NEPA only on paper will simply be too strong. Cf. Concerned Citizens on 1-190 v. Secretary of Transportation, 641 F.2d 1, 6 (1st Cir. 1981) (requiring a supplemental EIS when agency modifies plan so as to mitigate environmental harm would deter such commendable efforts).
For similar reasons, courts should be far less concerned with the form of an agency’s review of new information than with the substance. Whether the agency undertakes a full scale supplemental EIS process is less important than whether the agency gives the information any significant consideration. And as noted above, when the new information consists of a steady trickle of new scientific studies, it will often be impossible to determine when the information has become significant enough to require a supplemental EIS. See Friends of the River v. Federal Energy Regulatory Commission, 720 F.2d 93, 109 (D.C.Cir. 1983) (“Were we to order the Commission to reassess its decisions every time new forecasts were released, we would risk immobilizing the agency.”) The court should permit an agency to be flexible in considering new information relevant to ongoing projects so long as the agency shows that it has in fact given the information a hard look. See Warm Springs Dam, supra, 621 F.2d at 1025-26 (agency not required to formally supplement EIS when subsequent studies showed that agency had thoroughly considered new problem). As Judge Kennedy explained in his concurring opinion in Warm Springs Dam, courts should be satisfied with “other and less cumbersome ways for an agency to evaluate new information ____” 621 F.2d at 1027. Of course, when an agency has failed to consider new information, the supplemental EIS remedy will often be appropriate, as I think it was here. But if we demand more of agencies under threat of injunction in the first instance, we may, as a practical matter, achieve much less.
. Based on a surprisingly narrow and rigid reading of the complaint, the majority concludes that the district court should not have considered information that was not yet available at the time of the Navy’s 1981 decision to reactivate Project ELF. A fair reading of the complaint shows that it alleges a continuing breach of a continuing duty. Complaint f 19. The complaint alleges a violation of 40 C.F.R. § 1502.9(c), and that regulation is not limited in time — the agency must supplement the EIS at any time in the life of the project when significant new information comes to light. An agen*429cy breaches that duty when it goes forward with a project without considering new, relevant information within a reasonable time, and the breach continues until the agency considers the material. The plaintiffs here have alleged and proved precisely this sort of violation, and the district court was entitled to consider all new scientific information available up until a reasonable time before trial.
The majority’s overly narrow reading of the complaint is particularly odd because the majority concedes that the complaint could easily have been drawn to include information available both before and after the 1981 decision. Ante at n. 8. What the majority would accomplish with two narrower clauses, the plaintiffs accomplished with one broader one.
. In its review of the scientific evidence the majority engages in some interesting criticism of the scientific merits of plaintiffs’ evidence. However, even if the criticisms are valid, apart from their propriety on appellate review, the plaintiffs’ evidence still contrasts sharply with the Navy’s 1977 assertions that there was no "convincing,” "substantial" or "generally accepted” evidence of adverse biological effects from extremely low frequency radiation fields. See 578 F.Supp. at 1341-55, 1361-62.
. The Council on Environmental Quality adopted this regulation — which is binding on all federal agencies — as part of a comprehensive effort to improve and streamline NEPA procedures. The Department of Defense and the Navy have also expressly adopted this requirement in their own regulations. 32 C.F.R. § 775.3(b)(1) (1983); 32 C.F.R. § 214.6(D)(4) (1983).
. By holding that NEPA could have been violated only if the new information was itself significant enough to require a supplemental EIS, the majority conflicts with the Ninth Circuit’s decision in Warm Springs Dam.
. On motion for reconsideration, Judge Crabb identified several situations in which an injunction might not be necessary in a NEPA case, but this case does not fall within the exceptions she identified. Wisconsin v. Weinberger, 582 F.Supp. 1489, 1494-95 (W.D.Wis.1984).
. E.g., Concerned About Trident v. Rumsfeld, 555 F.2d 817, 823 (D.C.Cir.1976); 32 C.F.R. §§ 214.6(D)(4) and 775.3(b)(1) (1983) (Department of Defense and Navy regulations for NEPA compliance).