Every age has experienced scientific advances. A distinguishing feature of our era is the effort the scientific community is making to quantify the risks that seem inevitably to accompany the results of technological progress. The availability of this data has doubtless played a part in raising public consciousness about the mixed blessings of “progress,” and public concern has led to rigorous governmental regulation. These trends, in turn, have brought before the courts' controversies that present old issues in new contexts of unusual complexity. In determining whether regulatory actions conform to statutory requirements, courts are now obliged to review agency consideration of sophisticated data concerning the potential gravity of adverse consequences and the probability of their occurrence. This assessment of risk lies at the heart of this appeal, which involves a challenge to a federal regulation designed to reduce the risks from the transportation of radioactive materials. The United States Department of Transportation (“DOT” or “the Department”) appeals from the May 6, 1982, judgment of the District Court for the Southern District of New York (Abraham D. Sofaer, Judge) declaring invalid, in part, regulations promulgated by DOT to govern the transportation of large quantities of radioactive materials by highway. 539 F.Supp. 1237 (S.D.N.Y.1982). For reasons set out in detail below, we reverse and remand the matter for entry of a judgment upholding the regulations.
I.
The challenged DOT regulations can best be understood in light of developments preceding their issuance. In early 1976, the City of New York amended its health code to prohibit the transportation of spent nuclear fuel and other large quantities of radioactive material through the City without a Certificate of Emergency Transport from the Commissioner of Health. See N.Y. City Health Code § 175.111(7) (January .15, 1976). The amendment of the City’s Health Code effectively banned the use of motor vehicles to ship spent fuel from nuclear reactors operating on Long Island because all roads from Long Island pass through New York City. Since 1976, spent nuclear fuel has been removed from Long Island by barge across the Long Island Sound to New London, Connecticut.
The Brookhaven National Laboratories, which operates a reactor on Long Island, responded to New York City’s action by asking DOT to declare the amendment preempted by the Hazardous Materials Transportation Act (HMTA), 49 U.S.C. §§ 1801-*7371812 (1976 & Supp. V 1981).1 Section 105(a) of HMTA authorizes the Secretary of Transportation to promulgate rules governing the routing of shipments containing hazardous materials including radioactive materials. 49 U.S.C. § 1804(a). Section 112(a) of HMTA provides that any state or local regulation inconsistent with federal regulations is preempted. 49 U.S.C. § 1811(a).
In an “Inconsistency Ruling” published in April 1978, DOT found that HMTA did not preempt the amendment of the New York City Health Code. 43 Fed.Reg. 16,954 (Apr. 20, 1978). DOT reasoned that, although HMTA authorized the Secretary to develop national rules for the routing of nuclear materials, the Secretary had not yet exercised that authority. Consequently, in DOT’s view, municipalities like New York City were free to enact their own routing rules, including extreme routing requirements such as a ban on the shipment of nuclear materials through densely populated urban areas.2
As a consequence of restrictions placed on the transportation of nuclear materials by New York City and numerous other jurisdictions across the country, DOT decided to investigate whether federal rules governing highway carriers of radioactive materials might be needed. In an Advance Notice of Proposed Rulemaking published in August 1978, DOT expressed its concern that the various, inconsistent safety regulations imposed by state and local authorities might in fact diminish the overall safety of the transportation of nuclear materials. See 43 Fed.Reg. 36,492 (Aug. 17, 1978). In this Advance Notice, DOT outlined several proposals for regulating the highway transportation of radioactive materials at the federal level and invited comments on these proposals as well as on the threshold question, “Should radioactive materials be subject to more stringent Federal highway routing requirements than now imposed?” Id. at 36,-493-94.3 From the start, DOT indicated its intention to limit its proposed rulemaking to highway routing, which it identified as the mode of transportation that “offers the largest number of routing possibilities and the greatest access to population centers .... [while] fac[ing] immediate and significant disparities in safety requirements imposed by State and local jurisdictions.” Id. at 36,492.
In the course of rulemaking proceedings mandated by the Administrative Procedure Act, 5 U.S.C. §§ 551-706 (1976 & Supp. V 1981), DOT conducted eight public hearings and received more than 1600 comments on its proposed regulation of the highway transportation of radioactive materials. In January 1980, DOT published a Notice of Proposed Rulemaking, which contained DOT’s preliminary assessment of appropriate routing requirements and driver training programs. See 45 Fed.Reg. 7140 (Jan. 31, 1980). Almost a year later, DOT published a Final Rule, known as HM — 164, which closely resembled the proposed version. 46 Fed.Reg. 5298 (Jan. 19,1981) (codified at 49 C.F.R. §§ 171-173, 177 (1982)).
This litigation challenges the sections of the Final Rule governing the routing of motor vehicles that carry “large-quantity shipments” of radioactive materials.4 The Final Rule establishes a system of preferred routes comprising the highways of the Interstate Highway System supplemented by *738local highways selected and approved by state routing agencies.5 Under the Final Rule, vehicles carrying large-quantity shipments of radioactive materials should as a general matter “operate[] over preferred routes selected to reduce time in transit, except that an Interstate System bypass or beltway around a city shall be used when available.” 49 C.F.R. § 177.825(b) (1982). DOT designated the entire Interstate Highway System as a preferred route because of the System’s low accident rates and its capacity to reduce transit times. See 46 Fed. Reg. at 5300-01. However, because DOT believed that in many cases local roads might provide safer and more direct routes for highway carriers and that state authorities were better situated to determine where alternate routes would be preferable, state routing authorities were given authority to supplement the Interstate Highway System. See id. at 5301-02. The Rule also requires that large-quantity radioactive materials carriers prepare written route plans before shipment, that drivers for these shipments complete training programs, and that carriers moving irradiated reactor fuel follow security procedures established by the Nuclear Regulatory Commission (NRC). See 49 C.F.R. §§ 173.22, 177.825(c), (d), (e); see also 10 C.F.R. § 73 (NRC security regulations). Accompanying HM-164 was an appendix expressing DOT’s opinion that the Rule would preempt local regulations, such as the New York City Health Code, that “prohibit[ ] transportation of large quantity radioactive materials by highway between any two points without providing an alternate route for the duration of the prohibition.” 46 Fed.Reg. at 5317-18 (codified at 49 C.F.R. § 177 app. A (1982)).
In conjunction with the issuance of HM-164, DOT reckoned with section 102(2)(C) of the National Environmental Protection Act (NEPA), 42 U.S.C. § 4332(2)(c) (1976), which requires a detailed Environmental Impact Statement (EIS) in every proposal for a major federal action “significantly affecting” the quality of the human environment. DOT released with HM-164 a Final Regulatory Evaluation and Environmental Assessment,6 which concluded that an EIS was not needed because HM-164 would not have a “significant” impact on the environment. DOT’s Environmental Assessment noted that HM-164 would benefit the environment by reducing the existing level of background radiation. This benefit would result from a reduction in the transit time for shipments of radioactive materials. The Rule would also have a beneficial effect upon the risk that a shipment would be involved in a highway accident, since the interstate highway system enjoys a comparatively low accident rate.
However, the Environmental Assessment acknowledged that in one respect the Final Rule creates a risk. Permitting high-quantity shipments of spent nuclear fuel and other radioactive materials to be trucked through densely populated urban centers in jurisdictions that had previously banned such shipments creates an estimatable risk of serious consequences that would occur in the unlikely event of an accident with substantial leakage of radioactive gases. Conceding that this risk was credible, DOT concluded, on the basis of “overall risk” assessment, that this possibility did not cause HM-164 to have a “significant impact” on the environment. The concept of overall risk incorporates the significance of possible adverse consequences discounted by the improbability of their occurrence. DOT’s risk assessment, which we discuss in detail in Part IY, infra, relied on a study estimating the probability of a catastrophic accident resulting from the highway transportation of radioactive materials through urban centers to be approximately once every 300 million years. See Nuclear Regulatory Commission, Office of Standards Development, Final Environmental Statement *739on the Transportation of Radioactive Material by Air and Other Modes at v — vi (Dec. 1977) (hereinafter cited as 1977 NRC Report or NUREG-0170) (national estimate based on 1975 shipping rates); see also Sandia National Laboratories, Transportation of Radionuclides in Urban Environs: Draft Environmental Assessment 66 (July 1980) (hereinafter cited as Sandia Report) (study of probability of accident in New York City).
Throughout the rule-making process, the City of New York was a vocal critic of DOT’s proposals and repeatedly urged DOT to broaden the scope of its inquiry and to consider barging as an alternate means of transporting large-quantity shipments of radioactive materials around urban centers that lack circumferential highways. When the Department failed to incorporate the City’s barging suggestion into the Proposed Rule issued in early 1980, the City reiterated its support of barging and requested DOT to accompany the Final Rule with a non-preemption ruling for section 175.111(l) of the New York City Health Code. See HMTA § 112(b), 49 U.S.C. § 1811(b). The City’s application for a non-preemption ruling was denied as premature. Once the Final Rule was published in January 1981, the City renewed its application for a non-preemption ruling. On January 15, 1982, fifteen days before the Final Rule went into effect, DOT sent the City a letter indicating that the City’s non-preemption application would most likely not be approved for lack of substantial supporting documentation. The Final Rule went into effect on February 1, 1982.
However, orders entered in this litigation have prevented preemption of the City’s regulation. On March 25, 1981, the City filed a complaint in the Southern District of New York to invalidate Rule HM-164 on numerous grounds. Shortly after the complaint was filed, the District Court granted motions by the State of New York and a group of utilities to join the proceeding as intervening plaintiff and intervening defendants, respectively. On January 29, 1982, three days before HM-164 was to take effect, the District Court issued a temporary order restraining the preemptive effect of HM-164 upon section 175-111(7) of the City’s Health Code. On February 19, 1982, the District Court, in an exhaustive opinion, ruled that HM-164 violated HMTA and NEPA in its preemption of state and local bans on the transportation of large-quantity radioactive materials along highways in densely populated areas. The District Court permanently enjoined nationwide the enforcement of what it concluded was the invalid effect of HM-164, and offered the parties the opportunity to suggest corrections to the Court’s opinion. Based on these suggestions, the District Court filed an amended opinion and judgment on May 6, 1982, which limited its earlier decision and invalidated HM-164 only insofar as it preempted New York City’s Health Code. 539 F.Supp. 1237.
II.
We consider first the District Court’s interpretation of HMTA, the statute under which DOT issued Rule HM-164. The District Court construed HMTA to require that regulations promulgated under the statute set the safest feasible standards for transportation of hazardous materials. In the District Court’s view, HMTA regulations must maximize public safety not only for the nation as a whole, but for every local jurisdiction in the country as well. Furthermore, in the District Court’s view, DOT must require shippers to use the safest of alternative modes of transportation.7 Applying this construction of HMTA, the Court faulted DOT for not demonstrating that HM — 164 offers a safer means of transporting radioactive materials through New *740York City than barging these materials around the City. On this appeal, the Government and the intervening utility companies contend that HMTA does not require the Secretary to maximize safety, but rather authorizes her to set acceptable levels of safety for each mode of transportation. We agree with the appellants’ interpretation of the statute.
HMTA empowers the Secretary of Transportation “to protect the Nation adequately against the risks of life and property which are inherent in the transportation of hazardous materials in commerce.” HMTA § 102, 49 U.S.C. § 1801. The reference to “adequate” protection suggests that Congress expected the Secretary to exercise discretion in determining the appropriate level of safety. This understanding of HMTA is supported by section 105(a), which permits, but does not require, the Secretary to promulgate regulations governing “any safety aspect of the transportation of hazardous materials which the Secretary deems necessary or appropriate.” 49 U.S.C. § 1804(a).
A further clue that Congress did not intend HMTA regulations to maximize public safety, particularly on a jurisdiction-by-jurisdiction basis, is the statute’s regulatory structure. Under section 112(a), HMTA regulations preempt inconsistent state and local regulations. 49 U.S.C. § 1811(a). Congress included this provision “to preclude a multiplicity of State and local regulations and the potential for varying as well as conflicting regulations in the area of hazardous materials transportation.” S.Rep. No. 1192, 93d Cong., 2d Sess. 37 (1974) (hereinafter cited as Senate Report); see Kappelmann v. Delta Air Lines, Inc., 539 F.2d 165, 169-70 (D.C.Cir.1976), cert, denied, 429 U.S. 1061, 97 S.Ct. 784, 50 L.Ed.2d 776 (1977); Consolidated Rail Corp. v. City of Dover, 450 F.Supp. 966, 973-74 (D.Del.1978); see also National Tank Truck Carriers, Inc. v. City of New York, 677 F.2d 270, 275 (2d Cir.1982). To ameliorate the sweep of section 112(a), Congress wrote into HMTA a procedure whereby local jurisdictions could apply for non-preemption rulings for their own regulations. See HMTA § 112(b), 49 U.S.C. § 1811(b). A local jurisdiction qualifies for a non-preemption ruling if it can demonstrate to DOT that its local regulation “affords an equal or greater level of protection to the public” than that of DOT’s regulations, and “does not unreasonably burden commerce.” Id. This non-preemption procedure was added to HMTA so that in “certain exceptional circumstances” DOT could limit the preemptive force of federal regulations “to secure more stringent regulations” by local authorities. Senate Report, supra, at 38. It is difficult to understand why Congress would have added a non-preemption procedure to permit safer local regulations if the statute required the Secretary to adopt only those HMTA regulations that would maximize public safety on a jurisdiction-by-jurisdiction basis.
The District Court’s construction of HMTA would place tremendous, if not insuperable, constraints on the Secretary’s rule-making power in the absence of explicit statutory direction. Undoubtedly, Congress has (and occasionally uses) the authority to instruct a federal agency to regulate an area so as to maximize a public benefit or minimize a public harm. See, e.g., 33 U.S.C. § 1314(b)(2)(B) (1976) (“best measures and practices available to comply with” effluent limitations). But courts should not strain to infer from vague statutory language or legislative committee rhetoric a goal of maximizing a particular public policy. See FCC v. WNCN Listeners Guild, 450 U.S. 582, 593-95, 101 S.Ct. 1266, 1273-75, 67 L.Ed.2d 521 (1981); INS v. Wang, 450 U.S. 139, 144, 101 S.Ct. 1027, 1031, 67 L.Ed.2d 123 (1981) (per curiam). See Note, Intent, Clear Statements, and the Common Law: Statutory Interpretation in the Supreme Court, 95 Harv.L.Rev. 892, 910 — 12 (1982).
The District Court’s interpretation of HMTA would also require the Department to compare modes of transportation to determine which mode is the safest, thereby pitting the various modes of transportation in direct competition because only one mode of transportation can be the safest mode for *741any given, task. We do not believe that Congress imposed such a requirement.
Before HMTA was passed in 1974, a number of federal agencies, including the Federal Highway Administration and the Federal Railway Administration, supervised the transportation of hazardous materials. The Secretary of Transportation maintained a small technical staff to advise these agencies, see Hazardous Materials Transportation Control Act of 1970, Pub.L. No. 91-458, §§ 301-303, 84 Stat. 971, 977 (repealed 1974), but DOT had no substantive control over the various modes of transportation. Consequently, there was no coordinated regulation of the various modes.
In 1974, Congress passed HMTA to “draw[] the Federal Government’s now-fragmented regulatory and enforcement power over the movement of hazardous materials in commerce into one consolidated and coordinated effort under the direction of the Secretary of Transportation.” Senate Report, supra, at 1. For the first time, a single federal authority had the responsibility for overseeing the transportation of hazardous materials by all modes. This centralization was designed to achieve a comprehensive approach to reducing risk, for example, by ensuring that consistent packaging rules were enforced for all modes. See id. at 2.
Because HMTA consolidated federal regulatory control, the District Court inferred that the Act required the Secretary to permit the use of only the safest mode for any function. Such a requirement would constitute a radical shift in regulatory policy with serious ramifications for the transportation industry. In the past, we have been extremely reluctant to hold Congress to have made such a basic change in regulatory procedure absent explicit statutory language or other clear manifestations of congressional intent. Toilet Goods Ass’n v. Finch, 419 F.2d 21, 27 (2d Cir.1969). None of the provisions of HMTA remotely suggests that the Act was intended to mandate comparisons among modes of transportation. To the contrary, the House Committee made it clear that the Act would not change the Secretary’s regulatory authority. See H.Rep. No. 1083, 93d Cong., 2d Sess. 20 (1974), reprinted in 1974 U.S.Code Cong. & Ad.News 7669, 7680. We conclude, therefore, that in passing HMTA Congress intended DOT to use its centralized authority to develop a rational and consistent system of regulations for all modes of transportation, but expected that the Department would continue to follow the prior practice of regulating each mode of transportation on an individual basis.8
Having rejected the District Court’s construction of HMTA, we now reach the question of what standards HMTA sets to govern the Secretary’s rule-making authority. In our- view, HMTA regulations are valid as long as they are rationally related to the policy — the development of acceptable levels of public safety for each mode of transportation — underlying HMTA and are promulgated in accordance with the Administrative Procedure Act, 5 U.S.C. §§ 551-706 (1976 & Supp. V 1981). DOT fully complied with these standards. DOT followed appropriate notice and comment procedures, and the Final Rule was rationally related to the goal of developing acceptable levels of safety in the highway transportation of radioactive materials. Except for claims premised on an erroneous interpretation of HMTA, we do not understand the plaintiffs to dispute that HM-164 is a valid HMTA regulation.
III.
We next consider the District Court’s conclusions that DOT failed to comply with the requirements of NEPA. We discuss in this part the Court’s conclusion that by giving insufficient attention to barging as an alternative to HM-164 nationwide and for New York City in particular, DOT failed to consider alternatives to proposed agency action as required by NEPA. In Part IV, *742infra, we consider the Court’s ruling that DOT erred in assessing HM-164 not to have a “significant” environmental impact.
Section 102(2)(E) of NEPA obliges agencies to “study, develop, and describe appropriate alternatives to recommended courses of action in any proposal which involves unresolved conflicts concerning alternative uses of available resources.” The Secretary suggests preliminarily that HM-164 does not even encounter this requirement because it does not propose a “use” of a “resource” within the meaning of section 102(2)(E). This Court, however, has not construed section 102(2)(E) narrowly to apply only to agency actions that propose an identifiable use of a limited resource like park land or fresh water.9 Instead, we have ruled that federal agencies have a duty under NEPA to study alternatives to any actions that have an impact on the environment, even if the impact is not significant enough to require a full-scale EIS. Hanly v. Kleindienst, 471 F.2d 823, 834-36 (2d Cir.1972), cert, denied, 412 U.S. 908, 93 S.Ct. 2290, 36 L.Ed.2d 974 (1973); see Trinity Episcopal School Corp. v. Romney, 523 F.2d 88, 93 (2d Cir.1975).
HM-164 is an agency action with some impact on the environment. The Rule establishes the manner in which highly toxic substances will be moved around the country. Even apart from the risk of a possible accident, the permitted transportation will cause some contribution to the amount of low-level radiation on the interstate highway system. Under the standards developed in Hanly v. Kleindienst, supra, and Trinity Episcopal School Corp. v. Romney, supra, we conclude that HM-164 has sufficient impact oh the environment to require DOT to consider alternatives to the action.
This general conclusion, however, does not determine what sort of alternatives DOT had to consider in its Environmental Assessment and, in particular, whether the Department should have studied barging as an alternative to applying HM-164 to New York City.10 As the Supreme Court has recognized, an agency’s duty to consider alternatives under NEPA is “not self-defining.” See Vermont Yankee Nuclear Power Corp. v. National Resources Defense Council, Inc., 435 U.S. 519, 551, 98 S.Ct. 1197, 1215, 55 L.Ed.2d 460 (1978).
In the first instance, the agency itself is responsible for determining the range of alternatives to be considered, North Slope Borough v. Andrus, 642 F.2d 589, 601 (D.C.Cir.1980), and is supposed to follow what the District of Columbia has called a “rule of reason,” Natural Resources Defense Council, Inc. v. Morton, 458 F.2d 827, 834 (D.C.Cir.1972). An agency’s selection of alternatives, however, is not insulated from review. See County of Suffolk v. Secretary of Interior, 562 F.2d 1368, 1375 (2d Cir.1977), cert, denied, 434 U.S. 1064, 98 S.Ct. 1238, 55 L.Ed.2d 764 (1978). While an agency is not obliged to considér every alternative to every aspect of a proposed action, see Environmental Defense Fund, Inc. v. Corps of Engineers, 492 F.2d 1123, 1134-36 (5th Cir.1974); East 63rd Street Ass’n v. Coleman, 414 F.Supp. 1318, 1326 & n. 18 (S.D.N.Y.1976); Inman Park Restoration Inc. v. Urban Mass Transportation Administration, 414 F.Supp. 99, 115 (N.D.Ga.1976), aff’d sub nom. Save Our Sycamore v. Metropolitan Atlanta Rapid Transit Authority, 576 F.2d 573 (5th Cir.1978), reviewing courts have insisted that the agency “consider such alternatives to the proposed action as may partially or completely meet the proposal’s goal.” Natural Resources Defense Council, Inc. v. Callaway, 524 F.2d *74379, 93 (2d Cir.1975). Moreover, an agency will not be permitted to narrow the objective of its action artificially and thereby circumvent the requirement that relevant alternatives be considered. See id.; Committee to Stop Route 7 v. Volpe, 346 F.Supp. 731, 739-41 (D.Conn.1972).
The scope of alternatives to be considered is a function of how narrowly or broadly one views the objective of an agency’s proposed action. In this case, for example, if DOT’s objective is to improve the safety of highway transportation of radioactive materials, relevant alternatives might include a choice of routes, a choice of equipment, and a choice of driver qualifications. If DOT is concerned more broadly with all transportation of these materials, it might consider alternative modes of transportation. If the objective is viewed still more broadly as reducing the hazards of radiation exposure, the Department might consider alternative sources of power that could reduce the generation of spent nuclear fuel. Frequently, a pertinent guide for identifying an appropriate definition of an agency’s objective will be the legislative grant of power underlying the proposed action. See Picher, Alternatives Under NEPA: The Function of Objectives in an Environmental Impact Statement, 11 Harv.J. on Legis. 595 (1974); Comment, The National Environmental Protection Act of 1969: What “Alternatives” Must an Agency Discuss?, 12 Colum.J.L. & Soc.Probs. 221, 241-50 (1976). Statutory objectives provide a sensible compromise between unduly narrow objectives an agency might choose to identify to limit consideration of alternatives and hopelessly broad societal objectives that would unduly expand the range of relevant alternatives. We implicitly endorsed the pertinence of statutory objectives in ruling that an agency need not consider “alternatives which could only be implemented after significant changes in government policy or legislation.” Natural Resources Defense Council, Inc. v. Callaway, supra, 524 F.2d at 93.11
Having described the nature of an agency’s duty to consider alternatives, we now consider whether DOT satisfied that duty in this case. The Environmental Assessment prepared for HM-164 was a thirty-six page document supplemented by nine technical appendices. The Assessment covered nine alternatives, including a no-action alternative. All of the alternatives concerned highway transportation of radioactive materials. The two principal alternatives involved routing shipments around heavily populated areas, except for pickup and delivery, and, as HM-164 proposed in modified form, routing the materials via interstate highways, except for pickups and deliveries. The Assessment also considered requiring advance notice of large-quantity shipments as well as preparation of after-the-fact routing plans for these shipments. Finally, the Assessment considered time-of-day or time-of-week routing restrictions and specialized driver training programs for large-quantity shipments.
The District Court’s basic criticism of the Environmental Assessment was the absence of any serious consideration of the barging alternative. At various points in its opinion, the District Court seems to suggest that DOT should have considered barging as a nationwide alternative to trucking radioactive materials through densely populated areas. See, e.g., 539 F.Supp. at 1282. More often, however, Judge Sofaer makes clear that he was primarily concerned with DOT’s failure to consider barging as an alternative for New York City. See id. at 1281, 1287. This fundamental complaint was also directed at the Assessment’s treatment of the no-action alternative because, had DOT failed to promulgate HM-164, barging would have been the inevitable consequence in some jurisdictions, such as New York City. See id. at 1279.
*744Though we have required an agency to give some consideration to alternatives even though preparation of an EIS is not required, Hanly v. Kleindienst, supra, 471 F.2d at 834 — 36, it remains something of an anomaly to insist that an agency assess alternatives for an action that it has determined will not have a “significant” effect on the environment. See id. at 836-37 (Friendly, J., dissenting). But even accepting the teaching of Hanly, as we must, we are of the view that an agency’s finding of no significant impact, if otherwise valid, permits the agency to consider a narrower range of alternatives than it might be obliged to assess before undertaking action that would significantly affect the environment. In this case, DOT’s finding that HM-164 would not significantly affect the environment substantially diminishes the claim that the Department acted arbitrarily in declining to consider barging as a national alternative.
The claim is further undermined when we consider the statutory objective HM-164 was fulfilling. As we discussed in Part II, supra, HMTA authorizes the Secretary to promulgate regulations that establish acceptable national levels of public safety for each mode of transportation of hazardous materials. With HM-164, DOT consciously limited itself to studying and establishing national regulations for highway carriers, leaving for some future date the investigation of other modes of transportation. See 43 Fed.Reg. at 36,492. Since HMTA neither requires nor envisions that DOT will compare modes of transportation before issuing safety regulations, DOT was acting within its statutory mandate in limiting its inquiry to highway transportation. Even if barging offered a nationwide alternative to highway transportation, which seems unlikely,12 barging was not an alternative to DOT’s statutorily imposed objective of creating national safety regulations for highway transportation.
A third consideration that weighs heavily against requiring DOT to consider barging as a national alternative is the availability of the procedure for obtaining a non-preemption ruling. Of course, it was open to DOT to consider including in HM-164 a provision requiring (or at least permitting local preference for) barging in those circumstances where this alternative could safely and conveniently avoid highway transportation through urban centers. But the opportunity for a non-preemption ruling will enable the Department to give adequate consideration to the barging alternative in those areas where local circumstances might warrant it.
There is even less justification for requiring DOT to consider in its Environmental Assessment barging only around New York City. As a site-specific solution, barging around New York City offered no reasonable alternative to a nationwide regulation for highway transportation. Moreover, to fault DOT for not considering a barging alternative to New York City would imply that the Department should have made similar comparisons of transportation modes for numerous other jurisdictions. Enlarging the duty to consider alternatives to this degree would make it virtually impossible for an agency to promulgate national regulations like HM-164. See East 63rd Street Ass’n v. Coleman, supra; Inman Park Restoration, Inc. v. Urban Mass Transportation Administration, supra.
The availability of a non-preemption ruling is, in any event, a complete answer to the claim that DOT was required in its Environmental Assessment to consider the barging alternative specifically for New York City. When an agency undertakes a national regulatory scheme like HM — 164 and adopts procedures permitting localized environmental alternatives to be considered at a later date, we will not fault the agency for not considering these particular alternatives in its original Environmental Assessment. See County of Suffolk v. Secretary of Interior, supra, 562 F.2d at 1378.
*745In sum, we conclude that DOT did not violate NEPA by omitting from its Environmental Assessment consideration of barging as an alternative to HM-164 either for the nation as a whole or for New York City.
IV.
We turn now to the core issue: whether DOT violated NEPA by determining in its Environmental Assessment that HM-164 did not require preparation of an EIS because it was not an action that would “significantly” affect the environment. The District Court did not rule that an EIS was required. Instead, Judge Sofaer concluded that the Environmental Assessment was flawed in its consideration of whether HM — 164 would “significantly” affect the environment. Though he strongly intimated that DOT would be obliged to conclude that an EIS was required, once it remedied the shortcomings that he perceived in the Assessment, his ruling faulted the sufficiency of the Assessment, not the absence of an EIS. In Judge Sofaer’s view, the Assessment was deficient in its failure to reckon with numerous factors affecting the determination of how much risk HM-164 would create. Like Judge Sofaer, we approach the matter with an intuitive reaction that the transportation of radioactive materials through the cities of America poses risks that warrant careful consideration. However, after reviewing DOT’s analysis of the risks and the studies the Department relied upon, in light of the standards circumscribing our role, see Baltimore Gas & Electric Co. v. Natural Resources Defense Council, Inc., - U.S. -, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983), we conclude that the Department did not violate NEPA in deciding that an EIS was not required.
DOT undertook the Assessment to consider environmental ramifications of a national rule. The Assessment confronted the question of what effect HM-164 or possible alternatives to HM-164 would have on the environmental quality of the nation as a whole. The “no-action” alternative against which HM-164 and its alternatives were compared was not a national environment free of radioactive materials, but a national environment with radioactive materials and without comprehensive federal regulations governing their transportation by highway. DOT’s task in preparing the Environmental Assessment was complicated by the fact that the various options under consideration could not be tested against a single measure of environmental safety. In fact, for each alternative, the Department had to assess two distinct radiological risks: the effect of transporting radioactive materials on the level of background radiation and the risk that such transportation might result in accidents with the potential of causing serious damage.
Studies considered by DOT estimated the total annual dose of background radiation in the United States to be 40,000,000 rems. Of this total, transportation of radioactive materials, prior to HM-164, contributed 24,-360 rems, less than one-tenth of one percent of the total dose. One measure of the relatively slight health hazard presented by such transportation is the estimate in NU-REG-0170 that this background dose can be expected to cause an average of 3.07 latent cancer fatalities per year.13 DOT’s Environmental Assessment considered the estimated impact upon background radiation of two basic trucking alternatives: routing all large-quantity shipments of radioactive materials around heavily populated areas, except for pickups and deliveries, and routing such shipments on interstate highways, except for pickups and deliveries. The first option, avoiding population centers, was estimated by NUREG-0170, which DOT relied on, to reduce background radiation by 15.1 rems per year, which reduces latent cancer fatalities by an average of .002 per year. The “interstate” option, which HM-164 adopted in modified form, was estimated in the same study to reduce background *746radiation by 10.7 rems per year, which reduces latent cancer fatalities by an average of .0013 per year. None of the plaintiffs seriously claims that the impact of HM — 164 on background radiation levels — actually a slight improvement — constitutes a “significant” environmental effect.
What attracted the concern of the plaintiffs and of the District Judge was the risk that highway transportation of radioactive materials through urban centers might result in an accident of potentially serious dimensions. This possibility requires an analysis somewhat different from that undertaken in other NEPA cases. If the proposed agency action is the construction of a highway or the dredging of a harbor, some adverse environmental effects are certain to occur, such as destruction of park land or disturbance of animal or fish feeding grounds. Actions of this sort may also pose risks of other consequences that can only be estimated, but normally the seriousness of the certain consequences provides an adequate basis for determining that the effect on the environment will be sufficiently significant to require an EIS. In this case, however, the certain consequences are not at all significant. It is only the risk of accident that might render the proposed action environmentally significant. That circumstance obliges the agency to undertake risk assessment: an estimate of both the consequences that might occur and the probability of their occurrence.14
DOT’s Environmental Assessment faced up to this task. The Assessment initially noted the accident risk posed by highway transportation of radioactive materials by all modes of transportation in the absence of HM-164. The Assessment relied on the NUREG-0170 study, which used two measures of risk. First, the study estimated the average number of latent cancer deaths per year that could be expected from all types of accidents occurring during transportation of radioactive materials. The average figure was .017 deaths per year. Considerable perspective on this estimate is provided by comparing it to the 3.07 deaths per year estimated to result from the background radiation caused by transportation of radioactive materials. The background radiation risk is 181 times greater than the accident risk. The study’s second measure of risk estimated the probability of a transportation accident in any one year of sufficient seriousness to cause one or more early fatalities.15 The annual probability was 9.12 X 10 ”4. In other words, the Department estimated that an accident of sufficient seriousness to cause one or more early fatalities would occur approximately once every thousand years.
The Assessment then considered the impact upon accident risk of the two principal trucking alternatives for large-quantity shipments. The alternative of avoiding urban centers was estimated to cause a slight increase in the accident risk because this alternative entails circuitous travel on less *747safe secondary routes. The increase in latent cancer deaths from an accident was estimated to be .00031 per year and the increase in annual probability of an accident causing one or more early fatalities was estimated to be .37 X 10 -4. The interstate alternative, adopted by HM-164, was estimated to reduce latent cancer fatalities from an accident by .00062 per year and to decrease the annual probability of an accident causing one or more early fatalities by .9 X 10
The assessment also considered the consequences and probabilities of a “worst case” accident, relying on a study prepared for the NRC by Sandia Laboratories. See Sandia Report, supra. This study estimated the consequences of trucking accidents in New York City that would result in the maximum credible release of radioactive materials. The study estimated that the most lethal credible accident would be one involving plutonium shipments, which could cause 5 early fatalities, 1800 latent cancer fatalities, and 290 early morbidities (non-fatal injuries within a year). Id. at 66. The accident with the most severe economic impact would be one involving polonium, which could cause up to $9 billion in damages, primarily from reduced value of affected land. Id. However, notwithstanding the seriousness of these consequences, the Sandia Report concluded, “Since these accidents have very low probabilities of occurrence, they do not contribute significantly to expected values of risk.” Id. at 68. The NRC reached the same conclusion when it estimated that the probability of such a “worst case” accident occurring throughout the country in any year was 3 X 10 -9, based on 1975 shipment figures, approximately once every 300 million years. See NUREG-0170, supra, at v-vi.
Based on the Department’s overall risk assessment, the interstate alternative appeared to offer a cost-efficient solution to the transportation of large-quantity radioactive materials, achieving high levels of overall safety by requiring trucks to stay on the fast and relatively safe interstate highway system. However, the interstate alternative would permit trucks carrying large-quantity radioactive materials to travel through densely populated areas where a “worst case” accident with serious consequences was theoretically possible, although the probability was infinitesimal. Recognizing this theoretical risk, DOT modified the interstate alternative to require trucks carrying large-quantity shipments to use circumferential interstate highways around urban centers whenever possible. HM-164 also authorizes states to propose preferred local highways where interstate beltways are not available. In addition, the Rule requires drivers of large-quantity shipments to complete training courses designed to enhance safety and reduce the risk of accident. Having selected the interstate alternative with these modifications, DOT concluded, “The probability that a ‘worst case’ accident could occur is computed to be so low that when multiplied by even extraordinarily high consequences, the overall risk figure remains low,” in DOT’s view, so low that HM-164 would not have a “significant” impact on the human environment.
Though rejecting the adequacy of the Environmental Assessment, the District Court did not quarrel with DOT’s methodology. Judge Sofaer agreed that the Department was entitled to calculate overall risk by estimating possible consequences and then discounting them by the improbability of their occurring. However, he was critical of the manner in which the Department applied this methodology to the risk of high-consequence accidents. Judge Sofaer’s criticisms were three-fold: the Department had insufficiently analyzed the probability of a high-consequence accident; second, the Department had not fully discussed the consequences of such an accident; and finally, the Department had impermissibly considered the risk of a worst-case accident insignificant simply because the chances of such an accident are extremely low.
The issue we face is whether the District Court subjected the Environmental Assessment to a more exacting and intrusive review than is permissible under NEPA. NEPA establishes procedural re*748quirements designed to infuse environmental considerations into the Government’s decision-making processes. See Strycker’s Bay Neighborhood Council, Inc. v. Karlen, 444 U.S. 223, 227-28,100 S.Ct. 497, 499-500, 62 L.Ed.2d 433 (1980) (per curiam); County of Suffolk v. Secretary, supra, 562 F.2d at 1383. However, the Act mandates no particular substantive outcomes, and courts reviewing agency compliance with NEPA are limited to determining whether “the agency has taken a ‘hard look’ at the environmental consequences.” Kleppe v. Sierra Club, 427 U.S. 390, 410 n. 21, 96 S.Ct. 2718, 2730 n. 21, 49 L.Ed.2d 576 (1976). When an agency determines that a particular action will have no significant impact on the environment, a court may reject that determination only if it is arbitrary, capricious, or an abuse of discretion. See Morningside Renewal Council, Inc. v. United States Atomic Energy Commission, 482 F.2d 234, 238 (2d Cir.1973), cert, denied, 417 U.S. 951, 94 S.Ct. 3080, 41 L.Ed.2d 672 (1974); Hanly v. Kleindienst, supra, 471 F.2d at 829-30.
In its criticisms of DOT’s Environmental Assessment, the District Court took the Department to task for its treatment of highly technical matters relating to the probability, consequences, and overall risk of one particular facet of a rule that indisputably enhances safety on a nationwide basis. At this level of technical detail within a national rule-making process, errors in computation or shortcomings in analysis would have to be extraordinarily grievous to justify the District Court’s conclusion that the errors and omissions render the agency’s overall assessment arbitrary and capricious. Examining separately the District Court’s critique of DOT’s assessment of probabilities, consequences, and overall risk of high-consequence accidents, we see no legal infirmity in DOT’s conclusion.
A. Probabilities
The District Court did not disagree with the Department’s basic assessment of probabilities: “DOT’s conclusion that high-consequence accidents are very unlikely is supported by the record.” 539 F.Supp. at 1265. However, the Court faulted the Department’s probability assessment because it rested on several preliminary estimates that were themselves subjects of some dispute. For instance, experts disagreed about the number of large-quantity shipments that would occur in the future. In addition, there was some controversy over the Department’s assumptions about cask integrity during an accident. The District Court was also concerned that the Department did not fully consider the manner in which human error might affect its probability estimates. Finally, the Court criticized the Department for not explicitly dealing with the degree to which the risks of sabotage might affect the probability of a high-consequence accident.
Number of Shipments. Not surprisingly, experts disagree over how many shipments of large-quantity radioactive materials will be made in the future. In 1977, the NRC study known as NUREG-0170 predicted that there would be 1,911 shipments a year by 1985. See NUREG-0170, supra, at A-20 to -23; 539 F.Supp. at 1266. In 1979, another government report estimated that there would be no more than 641 such shipments in 1985. See Report to the President by the Interagency Review Group on Nuclear Waste Management (1979) (hereinafter cited as IRG Report). Other sources noted by the District Court indicate that as many as 2,500 shipments may be made annually in 1985. See 539 F.Supp. at 1267 & n. 9. Faced with this uncertainty, DOT based its calculations on the relatively conservative 1977 NRC study. At various points in the Environmental Assessment, the Department recalculated its projections using the lower IRG Report.
The District Court faulted DOT’s adoption of the NRC figures, supplemented by the IRG Report estimate, on the ground that the Department failed to discuss “the great uncertainties that exist on this subject and the possibility that the number of such shipments will increase exponentially.” Id. at 1267. Apparently, the District Court concluded that the uncertainties inherent in projecting shipment rates were sufficiently severe that the Department had to discuss *749the matter in its Environmental Assessment.16
We do not accept this conclusion. The Department conservatively selected a projection that was on the high side of the various estimates. The highest estimates for 1985 reveal shipment levels only slightly more than 25 percent greater than the ones accepted by the Department. Inasmuch as these projections would have to be off by several orders of magnitude before there would be any substantial impact on the probability of a high-consequence accident, the Department cannot be faulted for not discussing the relatively minor difference among experts in predicting 1985 shipment levels. Certainly the Department did not act arbitrarily in failing to engage in such a discussion.17
Cask Eeliability. The District Court also criticized the Department for commenting in its Advance Notice of Proposed Rulemaking that “packaging, in a severe transportation accident, would be expected to survive without any significant release of its contents.” 45 Fed.Reg. at 7143. While Judge Sofaer did not label this comment as arbitrary, see 539 F.Supp. at 1268-69, he noted that several commentators had found the risks of cask failure more severe, and he suggested that the Department should have discussed the conflicting evidence at greater length. Id.
We think that DOT’s assessment was well supported by two of the most complete studies on the transportation of radioactive materials. In particular, the NUREG-0170 study explicitly discussed cask reliability and explained why it found the risk of cask failure to be slight. See NUREG-0170, supra, at 5-20 to —21. As the District Court noted, in light of this scientific support, the Department’s conclusion as to cask reliability was not arbitrary. Accordingly, it should not be susceptible to attack on judicial review.
Human Error. The District Court also faulted DOT for relying on probability assessments that “fail to take into account the potential effect of human error.” 539 F.Supp. at 1269. This objection stems from the Court’s concern that the Department unreasonably underestimated the likelihood of high-consequence accidents by assuming that every shipment of large-quantity radioactive materials would be packaged and handled without human error. The Court apparently feared that inevitable human mistakes would make large-quantity shipments more susceptible to leakage and contamination during an accident than the Department had assumed. Id.
This criticism is unjustified. Both the NUREG-0170 study and the 1980 Sandia Report analyzed the impact of human error. As the District Court noted, the Sandia Report devoted a chapter to the subject, see Sandia Report, supra, at 71-84, and the NRC also discussed the matter under the heading of “Abnormal Transport Occurrences,” NUREG-0170, supra, at 4-31 to -33. The NRC report indicated that human errors compromise cask integrity in only a small fraction of shipments. Since large-quantity shipments are packed in such massive containers, human error is even less likely to affect overall package integrity of these shipments. Id. at 4-44 to -48. The Sandia Report agreed that “[m]ost human errors do not produce compromises of package integrity.” Sandia Report, supra, at 15. In light of the extremely low probability that any cask involved in a truck accident would also be compromised by human error, it does not seem to be reasonable to criticize *750DOT for not factoring this contingency into its calculations. Moreover, the Department corroborated its probability estimates by looking to the historic accident rates. Since the effect of human error is inevitably included in the historic rate of accidents, the Department, at least implicitly, considered human error in its probability assessment. For a source of risk this insignificant, implicit consideration was adequate.
Sabotage Risks. In its advance Notice of Proposed Rulemaking, DOT acknowledged that many observers had expressed concern over the possibility that saboteurs or terrorists might disrupt nuclear shipments and precipitate high-consequence accidents. The Department declined to assess this possibility on jurisdictional grounds:
Development of the current DOT proposal reflects existing arrangements between DOT and NRC wherein NRC exercises responsibility for any necessary physical security requirements during transportation. The DOT proposal is therefore directed at reducing impacts associated with normal and accident situations arising in transportation, while NRC is concerned with preventing malicious or deliberate release of radioactive materials.
45 Fed.Reg. at 7144. In 1979, the NRC had promulgated interim security rules governing the transportation of irradiated fuels, see 10 C.F.R. §§ 73.25-.37 (1982). The Department’s sole response to the threat of sabotage was to propose extending the NRC regulations to shippers outside the NRC’s jurisdiction.
The District Court had no complaint with DOT’s decision to defer to the NRC as the appropriate agency for supervising matters of physical security. Nevertheless, Judge Sofaer felt that the Department should have included a discussion of sabotage in its assessment of the probability of high-consequence accidents. Under regulations of the Council on Environmental Quality, agencies should consider “[t]he degree to which the possible effects on the human environment are highly uncertain or involve unique or known risks.” 40 C.F.R. § 1508.27(b)(5) (1982) ; see DOT Order 5610.1C (Sept. 18, 1979). Both the Sandia Report and the NRC have accepted that sabotage presents a real although unquantifiable risk.for the transportation of large-quantity radioactive materials. See Sandia Report, supra, at 85 (“Sabotage involves human motivations and the probability of human actions which are unquantifiable with our present knowledge.”); 45 Fed.Reg. 37,402 (1980) (“[Estimates of the probability of successful sabotage of spent fuel shipments cannot be made with any confidence.”). However, other experts have insisted that sabotage-induced, high-consequence accidents are “essentially impossible.” NUREG-0170, supra, at 7-2. The District Court concluded, “[G]iven the conflicts in the underlying data, [DOT] was obligated to state its view on the probability of such an event, even if that view was only that no estimate could reasonably be made.” 539 F.Supp. at 1271.
With respect to environmental consequences that are only remote possibilities, an agency must be given some latitude to decide what sorts of risks it will assess. See Natural Resources Defense Council, Inc. v. NRC, 685 F.2d 459, 516, 540-45 (D.C. Cir.1982) (Wilkey, J., dissenting), rev’d,U.S. -, 103 S.Ct. 2246, 76 L.Ed.2d 437 (1983) . Here DOT simply concluded that the risks of sabotage were too far afield for consideration. To a large degree this judgment was justified by the record. Substantial evidence indicated that sabotage added nothing to the risk of high-consequence accidents. Even the least sanguine commentators could say only that sabotage added an unascertainable risk. In light of these conflicting points of view, it was within DOT’s discretion not to discuss the matter further beyond adopting the NRC security requirements.
B. Consequences
Physical Consequences. Despite the discussion of high-consequence accidents in the Environmental Assessment and the Department’s references to more detailed analyses of the subject, the District *751Court found that the Department had treated the subject arbitrarily: “[I]n over a period of about five years, estimates of the consequences of the most severe accident have escalated dramatically. DOT must address these uncertainties and their implications for rational planning.” 539 F.Supp. at 1273.
Here again, the District Court’s demands on DOT’s Environmental Assessment were excessive. The Department cited and commented upon the. Sandia Report’s estimate of the extent of a high-consequence accident. The Sandia Report’s estimate was toward the high end of the scale of predictions offered to the Department.18 For a freakish accident that is only theoretically possible, the Department did not commit reversible error by conservatively accepting an estimate of serious impact without noting and attempting to analyze professional disagreement over the precision of the estimate.
Social Consequences. The District Court also faulted DOT’s Environmental Assessment for its failure to consider the psychological impact that HM-164 would have on the general public, even in the absence of any serious accidents. Since the District Court wrote its opinion, the Supreme Court has determined that fear is not a cognizable environmental impact under NEPA. See Metropolitan Edison Co. v. People Against Nuclear Energy, -U.S. -, 103 S.Ct. 1556, 75 L.Ed.2d 534 (1983).
C. Overall Impact
At the heart of the District Court’s criticisms of the Environmental Assessment lay its dissatisfaction with DOT’s use of the results of the assessment of overall risk. In Judge Sofaer’s view,
DOT’s use of overall risk obfuscated the central question as to the environmental significance of its proposed action: whether a proposal entailing a credible risk of catastrophe demands treatment as one “significantly affecting the quality of the human environment.”
539 F.Supp. at 1274 (quoting 42 U.S.C. § 4882(2)(C)).
This final criticism can be understood in two ways, neither of which justifies declaring the Department’s action arbitrary or capricious. First, Judge Sofaer could have meant that DOT was required to use some other form of risk assessment than overall risk analysis. Certainly, other risk techniques were available, and Judge Sofaer apparently preferred the so-called “mini-max” methodology employed by the Sandia Report.19 Id. at 1275. However, in reviewing an environmental assessment, a district court is precluded from imposing its own choice of risk analysis upon a federal agency. As long as the agency’s choice of methodology is justifiable in light of current scientific thought, a reviewing court must accept that choice. In this case, DOT had discretion to apply overall risk analysis in its Environmental Assessment.
Alternatively, Judge Sofaer might have meant' that DOT was wrong to conclude that HM-164 would have no significant impact on the environment once the Department had conceded the infinitesimal probability that the Rule might lead to catastrophic consequences. We believe, especially in light of the Supreme Court’s recent caution in Baltimore Gas, supra, that this view does not comport with the degree of deference that courts should accord to agency determinations concerning environmental consequences. In Kleppe v. Sierra Club, supra, 427 U.S. at 410 n. 21, 96 S.Ct. *752at 2730 n. 21, the Supreme Court stated that neither NEPA nor its legislative history “contemplates that a court should substitute its judgment for that of the agency as to the environmental consequences of its actions.... The only role for a court is to insure that that agency has taken a ‘hard look’ at environmental consequences.” Here, DOT considered a rule that might be expected to generate a catastrophic accident approximately once every 300 million years. After receiving advice from all sides, the Department decided that such a remote possibility, even of a serious consequence, did not create a “significant” risk for the human environment. Disquieting as it may be even to contemplate such matters, this decision cannot be said to be an abuse of discretion.20
Y.
Throughout its opinion, the District Court emphasized the effect of HM-164 on New York City. To a large extent the proceedings in the District Court can be viewed not as a review of the validity of HM-164 as applied to New York City but as an inquiry into DOT’s anticipated denial of the City’s request for a non-preemption ruling under section 112(b) of HMTA. To the extent that this occurred, it was premature. In framing HMTA, Congress decided that federal regulations would presumptively preempt inconsistent local regulations and that the local authorities would then have the burden of demonstrating to DOT that their local regulations provided greater safety without burdening interstate commerce. Courts are not free to reverse this presumption or to shift the burden of proof from state to federal authorities.
If DOT ultimately denies New York City’s pending application for a non-preemption ruling, the City can seek judicial review of that denial. At that time, a federal court will review the Department’s action, if adverse, with the benefit of both the Department’s explanation for its denial and a fully developed record comparing New York City’s Health Code to HM-164 as applied to New York City. If the Court determines that the Department impermissibly denied the City’s application for non-preemption, adequate relief will then be available.21 In the interim, however, Congress has legislated that the federal rule will govern.
The judgment of the District Court is reversed, and the case is remanded with instructions to enter a judgment upholding HM-164.
. Initially, the United States challenged the amendment to the New York City Health Code in federal court. United States v. City of New York, 76 Civ. 273 (IBW) (S.D.N.Y. Jan. 15, 1976). After the District Court denied the Government’s motion for a preliminary injunction, the case was transferred to the suspense calendar by stipulation.
. The Inconsistency Ruling concluded that, if DOT chose to establish national routing procedures in the future, the New York City ordinance might become preempted unless the City received a non-preemption ruling from DOT, as permitted by section 112(b) of HMTA, 49 U.S.C. § 1811(b).
. At that time, federal regulations simply stated that unless impracticable, carriers of hazardous materials should use routes that avoid heavily populated areas. 49 C.F.R. § 397.9 (1978); see 43 Fed.Reg. at 36,492-93.
. “Large quantity radioactive materials” are defined in 49 C.F.R. § 173.389(b) (1982).
. See “Guidelines for Selecting Preferred Highway Routes for Shipments of Large Quantity Radioactive Materials,” 49 C.F.R. § 277.-825(b)(1)(ii) (providing directions for state routing agencies).
. The Department had prepared a draft Environmental Assessment to accompany its Notice of Proposed Rulemaking. See 45 Fed.Reg. at 7152 (summary of draft assessment).
. The District Court viewed the Act as “a simple yet strong direction” for the Secretary “not only to consider alternatives to proposed regulations but also, when considering alternatives, to adopt the safer option unless it is impractical.” 539 F.Supp. at 1289; see id. at 1290. From this mandate, the District Court concluded that it was impermissible for the Secretary to promulgate HMTA regulations that “declare a certain level of safety ‘acceptable’ regardless of the possibility of achieving higher levels through reasonable measures.” Id. at 1289.
. In so ruling, we do not mean to imply that DOT cannot prohibit the transportation of certain classes of hazardous materials by one particular mode if the Department concludes that there is no practical means of regulating that mode to ensure án acceptable level of safety.
. Arguably, HM-164 satisfies the literal meaning of section 102(2)(E) since the movement of radioactive materials along highways can be considered a “use” of the surrounding air.
. Two different NEPA provisions deal with the analysis of alternatives. Section 102(2)(E), 42 U.S.C. § 4332(2)(E), is the more general provision, applicable even when an agency need not prepare a complete EIS. See Hanly v. Kleindienst, supra, 471 F.2d at 834-35 (discussing previously numbered section 102(2)(D)). Section 102(2)(C)(iii), 42 U.S.C. § 4332(2)(C) (iii), requires a more detailed statement on alternatives to the proposed action, but only applies when an EIS is required.
. This view is at odds with cases such as Natural Resources Defense Council, Inc. v. Morton, supra, in which the District of Columbia Circuit set aside a rule governing natural gas exploration because the rule’s EIS did not consider energy conservation as an alternative. The Court viewed conservation as a cognizable alternative to the societal objective behind the agency’s action — meeting the nation’s energy requirements.
. Twenty-six percent of the nation’s nuclear facilities are reportedly not directly accessible to navigable waters. See NUREG-0170, supra, at 6-9.
. A latent cancer fatality is a death that results from exposure to radiation, that occurs more than a year after the exposure, and that would not have occurred in the absence of the exposure.
. We recognize that in a sense all evaluations mandated by NEPA are predictions about the future. The Act is concerned with the future effects of proposed agency actions, Scientists’ Institute for Public information, Inc. v. Atomic Energy Commission, 481 F.2d 1079, 1088-89 (D.C.Cir.1973); Momingside Renewal Council, Inc. v. United States Atomic Energy Commission, 482 F.2d 234, 241-42 (2d Cir.1973) (Oakes, J., dissenting), cert, denied, 417 U.S. 951, 94 S.Ct. 3080, 41 L.Ed.2d 672 (1974). Nevertheless, there is a difference between assessing the future consequences of agency actions that “will affect the environment,” Scientists’ Institute, supra, 481 F.2d at 1089, and assessing the risk that proposed actions might have environmental effects, especially in the event of accidents. The fact that effects are only a possibility does not insulate the proposed action from consideration under NEPA, but it does accord an agency some latitude in determining whether the risk is sufficient to require preparation of an EIS. “[A] reviewing court must remember that the Commission is making predictions, within its area of special expertise, at the frontiers of science. When examining this kind of scientific determination, as opposed to simple findings of fact, a reviewing court must generally be at its most deferential.” Baltimore Gas & Electric Co. v. National Resources Defense Council, Inc., supra, - U.S. at-, 103 S.Ct. at 2256.
. An early fatality is a death that results from radiation exposure, that occurs soon after the exposure or at least within one year of the exposure, and that would not have occurred in the absence of the exposure.
. In a footnote, the District Court contends that DOT erred in interpreting projections in the IRG Report. See 539 F.Supp. at 1267 n. 8. Even if this criticism were well taken, it would not affect the Department’s Assessment, which was based on the more conservative figures developed in the 1977 NRC study.
. The District Court also faults DOT for using historical accident rates to corroborate its projections. See 539 F.Supp. at 1267-68. The Court felt that the use of these rates was improper because, elsewhere, the Department had conceded that the figures were not wholly reliable. While it might have been arbitrary for the Department to base its entire analysis on statistics of doubtful validity, it was not improper for it to introduce these figures to double check estimates derived through independent means.
. Other commentators predicted that sabotage could exact a higher toll in human life than the Sandia Report estimated, see 539 F.Supp. at 1272, but the Sandia Report was generally more conservative than other studies. For instance, the NRC itself estimated that a “worst case” accident would cause 174 latent cancer fatalities with $800 million of economic damage, whereas the Sandia Report estimated that latent cancer fatalities could run as high as 1800 and economic damages as much as $9 billion. Compare NUREG-0170, supra, at 5-38 to -53, with Sandia Report, supra, at 66.
. Under “mini-max” analysis, the Sandia Report simply analyzed the worst possible consequences of a highway accident without considering the likelihood of that event.
. Our dissenting colleague appears to take the view that the very existence of the “worst case” possibility would be sufficient to require preparation of an EIS, regardless of the infinitesimal probability that the “worst case” accident will happen. We do not doubt the general proposition that “worst cases” do occur. Planes crash, and the Titanic sank. What we reject is an automatic rule requiring preparation of an EIS for every action that has any possibility, however remote, of causing serious accidental injury. Such a rule would routinely require an EIS for federal actions, since it is hard to imagine any agency action involving people or equipment that is not subject to some estimatable risk of causing serious accidental injury. In this case, reasonable minds may differ as to whether or not the gravity of the “worst case” accident or other less serious accidents, discounted by their improbability, presents an overall risk of sufficient significance to warrant an EIS. So long as that choice falls within the range of reasonable dispute, an agency’s informed decision not to require an EIS is neither arbitrary nor capricious.
. The parties differ over whether New York City would be entitled to a non-preemption ruling if it can show that its local regulations enhance safety without burdening interstate commerce or whether DOT retains discretion to deny a non-preemption ruling even if the City makes such a showing. This question is not before us, and we do not reach it.