concurring in part and dissenting in part:
The FAA has determined that the progressive anatomic, physiological, and cognitive decline generally associated with aging means that all pilots over the age of 60 represent too great a threat to aviation safety to be allowed to fly in part 121 operations, which constitute by far the bulk of commercial common carrier operations.1 The FAA based this determination not on evidence demonstrating that pilots over 60 perform less well than pilots under 60, but rather on the claim that there is no accurate means to identify which pilots are particularly at risk of suffering a sudden incapacitation or more subtle deterioration in their abilities and the age of 60 was within the age range where the incidence of diseases associated with aging sharply increases. This argument is essentially the same as that which the FAA offered when it first adopted the Age 60 Rule in 1959. The agency continued to adhere to the position that the Age 60 Rule is necessary to ensure the highest level of aviation safety despite the medical and technological developments over the ensuing nearly four decades, despite a growing trend among foreign aviation authorities to allow pilots over 60 to fly, and despite a recent report commissioned by the FAA which concluded there was no evidence of an increase in accidents associated with older pilots at least up to age 63.
The majority concludes that this decision by the FAA does not violate the Age Discrimination in Employment Act (“ADEA”) and satisfies the Administrative Procedure Act’s (“APA”) prohibition on arbitrary and capricious agency action. I agree that the ADEA does not directly govern the FAA in its role as a regulator, and thus the FAA need not prove that the Age 60 Rule is a bona-fide occupational qualification for pilots. But I believe that the FAA’s justification for the rule simply does not pass muster under the APA.2 It may be the case that our current medical knowledge and testing protocols are unable to identify those older pilots who are at risk of sudden incapacitation or subtle deterioration in functioning, so that an arbitrary across-the-board age limit remains the only reliable means of achieving the highest possible level of aviation safety. However, the FAA has not yet provided an adequate justification on this go-round for its conclusion that this situation still exists, nor for its determination that aviation safety requires all common carrier pilots, even those carrying cargo only, to be subject to the age limit but not corporate or air-taxi pilots. The FAA’s decision also suffers from a reliance on flawed and inapplicable studies of accident *771rates. Perhaps hardest to swallow is the FAA’s continued refusal to try to obtain medical or performance data on older pilots at the same time as it claims that such evidence is required before any change in the rule can be countenanced. The agency’s complacent acceptance of this Catch-22 situation, particularly given that the result is the continuation of a government-imposed regime of age discrimination, seems to me the epitome of arbitrary action.
I. The FAA’s Failure to Offer an Adequate Explanation of the Need for the Age 60 Rule
The core of the APA’s prohibition on arbitrary and capricious agency action is the requirement that an agency must provide a reasoned explanation for what it does. While “[t]he scope of review under the ‘arbitrary and capricious’ standard is narrow and a court is not to substitute its judgment for that of the agency,” neither may a court sanction agency action when the agency merely offers conclusory and unsupported postulations in defense of its decisions or when it ignores contradictory evidence in the record and fails to justify seeming inconsistencies in its approach. Motor Vehicle Mfrs. Ass’n v. State Farm, Mut. Auto. Ins. Co., 463 U.S. 29, 41, 103 S.Ct. 2856, 2865-66, 77 L.Ed.2d 443 (1983); accord Dickson v. Secretary of Defense, 68 F.3d 1396, 1404-05 (D.C.Cir.1995). The FAA’s explanation of its decision to retain the Age 60 Rule suffers from all of these defects.
The FAA currently grants medical exemptions to pilots under 60 who are at risk of sudden incapacitation or subtle deterioration in functioning because of known medical conditions, but refuses to grant exemptions to pilots over the age of 60 who are at risk of these same effects because of aging. In its decision, the FAA argued that this differential treatment of younger pilots and older pilots is merely a reflection of the state of medical technology; according to the FAA, there are tests by which the status of a known cardiovascular or neurological condition can be reliably monitored, but there are no tests by which the presence of such a condition can be reliably determined. The majority accepts this claim as an established fact, but the FAA cites no evidence in its support and it certainly is not intuitively clear why the difference between being at risk of sudden incapacitation or subtle deterioration because of known medical conditions and being at risk of the same effects because of aging should produce radically different diagnostic capacities. Cf. Baker v. FAA, 917 F.2d 318, 321 (7th Cir.1990); see also id. at 325 (Will, J-, dissenting).3 If the difficulty in the latter instance is the lack of certainty regarding the pilot’s health, why not simply presume that the older pilot does in fact suffer from conditions that could cause these effects and then determine whether the pilot is safe to fly by means of the same tests that *772are used to monitor pilots with known conditions?
Nor can the FAA’s differential treatment of older and younger pilots be explained on the grounds that the age-related conditions that pose a concern in regard to older pilots are different from those found in younger pilots. The record makes clear that the FAA allows younger pilots to fly who have been diagnosed with cardiovascular disease, even though it also lists cardiovascular disease as one of the potential causes of sudden incapacitation and subtle deterioration in older pilots. Age 60 Rule, 60 Fed.Reg. at 65,983-84. In any event, it is the potential effects of the medical conditions associated with aging, and not the conditions per se, that pose a risk to aviation safety, and the FAA allows younger pilots to fly with conditions that could produce the same effects. For example, while the FAA puts great emphasis on the subtle decline in cognitive functioning that accompanies aging, the FAA allows younger pilots to fly with various conditions, such as neurological and psychological disorders or alcoholism, that potentially could undermine a pilot’s cognitive abilities. J.A. 206, 284, 611-12, 678. The FAA nowhere explains why the same tests that it employs to assess whether younger pilots with these conditions possess the level of cognitive functioning required to fly safely cannot be used to make the same determination regarding older pilots.4
The FAA’s failure to explain adequately its different treatment of younger pilots and older pilots when both are at risk of sudden incapacitation or subtle deterioration in functioning is not simply a minor deficiency in its analysis. The heart of the FAA’s defense of the Age 60 Rule is its claim that medical knowledge does not provide a means by which those pilots at risk of these effects can be accurately identified. If the tests that the FAA currently uses to monitor younger pilots with known conditions that might cause such effects could be used to monitor older pilots at risk of these conditions, then there would be no need for the rule.
A similar lack of reasoned analysis characterizes the FAA’s explanation of the scope of the Age 60 Rule. The rule currently applies to all part 121 pilots, including part 121 pilots that work solely in cargo operations. It also applies only to part 121 pilots; other pilots— such as corporate pilots, test pilots, inspectors, and air-taxi pilots—are not covered. The FAA, defending its application of the Age 60 Rule to all and only part 121 operations, noted simply that Congress required the FAA in promulgating safety regulations to consider the duty of air carriers, which transport passengers or property by aircraft as a common carrier, to perform their services with the highest possible degree of safety in the public interest and to take account of the differences between air carriers and other forms of air commerce. See Age 60 Rule, 60 Fed.Reg. at 65,985. According to the majority, this statement suffices to explain the FAA’s refusal to apply the Age 60 Rule to non-air carrier pilots, such as corporate pilots, because “Congress clearly left the FAA free to regulate corporate aircraft operations at less than the highest possible degree of safety.” Op. at 768. But even if Congress did not require the highest possible degree of safety for non-air carrier operations, it did authorize the FAA to prescribe regulations that the FAA “finds necessary for safety in air commerce operations.” 49 U.S.C. § 44701(a)(5) (1994). Thus, what is needed before we can conclude that the *773FAA’s exclusion of non-air carrier operations from the Age 60 Rule is reasonable is some explanation as to why the rule is not required to achieve the level of safety that the FAA believes is appropriate for non-air carrier operations. Such an explanation is not to be found anywhere in the FAA’s decision.
Further, the FAA’s citation of the duty of air carriers to operate with the highest possible degree of safety does not explain why the FAA applies the Age 60 Rule to some but not all air carriers. At the same time as it issued its decision refusing to rescind the Age 60 Rule, the FAA also promulgated a final regulation extending the rule to commuter operations as part of a shift of most commuter airlines to part 121. However, as applied to commuter operations the Age 60 Rule will not be fully effective for four years. The FAA justified this four year delay in terms of the burden the rule would impose on commuter airlines and commuter pilots; it emphasized that commuter airlines have invested money in training pilots in the expectation they could fly past 60 and will be subject to numerous other new regulations in the immediate future as a result of the shift to part 121, while commuter pilots have not planned on leaving their positions at 60. These are all sensible reasons to delay enforcement of the Age 60 Rule, but they are difficult to reconcile with the FAA’s citation of the duty of air carriers to perform with the highest degree of safety in defending the application of the rule to all part 121 operations. Why, if this duty does not allow the FAA room to consider the differences between cargo and passenger operations or the burden on pilots under part 121, does it allow the FAA to take economic and fairness concerns into account in regard to commuter operations?
The FAA also provides no explanation for why duty to perform with the highest degree of safety possible is consistent with the continued exclusion from the Age 60 Rule of air-taxis, which are air carriers but not regulated under part 121. The FAA’s only reference to air-taxis comes not in its age 60 decision but in its proposal to shift commuter operations to part 121, and consists merely of the statement that air-taxis “are unlike commuter or major air carrier operations,” along with the comment that the FAA was only asked to put commuter operations under part 121. See Commuter Operations and General Certification and Operations Requirements, 60 Fed.Reg. 16,230, 16,235 (1995). The majority maintains that the exclusion of air-taxi operations from the Age 60 Rule is reasonable simply because the FAA was not asked to extend the rule to such operations. I find this argument wholly unpersuasive and wide of the mark. The question is not whether the rule should be extended to air-taxi operations, but rather whether the FAA’s application of the rule to all air carrier operations coming under part 121 was reasonable. Given that its defense of the application of the Age 60 Rule to all part 121 operations was simply a citation of the air carrier’s heavy duty of safety, the FAA has an obligation to provide some explanation as to why this duty does not require non-part 121 air carrier operations to be subject to the Age 60 Rule as well.5
A third deficiency in the FAA’s decision is its reliance on flawed and inapplicable studies of accident rates. As the majority details, one instigating factor behind petitioners’ request for a rulemaking on the Age 60 Rule was the 1993 Hilton report on the relationship between pilot age and accident rates. This report, commissioned by the FAA, determined that there was no evidence of increased accident rates for air carrier pilots as they neared 60, and if anything a slight downward trend. Based on its analysis of pilots over the age of 60 flying in operations not covered by the Age 60 Rule, the report concluded that the age limit for part 121 pilots could be extended from 60 to 63. The *774FAA. concluded, however, that the Hilton report did not justify rescission of the Age 60 Rule or an increase in the age cutoff.
The FAA based its rejection of the Hilton report in large part on the fact that earlier studies had found an increase in accident rates. But the record indicates that many of these studies are seriously flawed. For example, the FAA described in great detail the conclusions of a 1983 report that found a substantially higher accident rate for pilots over 60 than for younger pilots. While the FAA noted that several commentators disagreed with the methodology used in this report, it never explained why, despite these methodological problems, the 1983 report retained any evidentiary value. As the Seventh Circuit commented, these problems— such as including millions of nearly accident free air carrier miles in determining the accident rate for pilots under 60 even though as a result of the Age 60 Rule no air carrier miles are available in determining the accident rate for pilots over 60, and calculating a single accident rate for all pilots in a ten year cohort—mean that the 1981 report is, at best, of “very limited usefulness.” Baker, 917 F.2d at 320-21 & n. 1; accord J.A. at 100-01, 659; see also J.A. at 101-02, 500-01, 518-20 (criticisms of the Office of Technology Assessment report and the second Golasewski report). The FAA also relies on a National Research Council report that found car accident rates increased for older drivers, arguing that the lack of part 121 pilots flying past 60 necessitated the use of surrogate data to assess the performance of older pilots and that car accident data was a relevant comparison because both flying and driving a car require good reflexes and judgment. But, again as the Seventh Circuit has noted, “[t]he connection between automobile drivers and pilots itself seems tenuous given the pilots’ training, demonstrated proficiency, medical fitness, etc.” Baker, 917 F.2d at 321; see also J.A. at 504. The FAA does not acknowledge these differences or explain why, even so, car accident data is relevant.
II. The Need for Evidence on Older Pilot Safety and the FAA’s Refusal to Obtain it
These deficiencies in the FAA’s justifications for the Age 60 Rule lead me to conclude that the FAA’s decision to retain the rule fails the reasoned decisionmaking requirements of the APA. Even without them, I believe that the FAA’s refusal to try and obtain the evidence it claims is necessary to rescind the rule would require us to hold that its decision was arbitrary.
In its decision, the FAA emphasizes that what is required before a conclusive assessment of the contribution of the Age 60 Rule to protect aviation safety can be reached is medical and performance data on pilots over 60 serving in air carrier operations. The importance of such data was also underscored by the National Institutes of Health (“NIH”) in its 1981 report on the Age 60 Rule, and the need for such directly relevant data is hard to deny. But the FAA simultaneously refuses to allow any over-60 pilots whose performance and health could be studied to generate this data to fly in air carrier operations, even pilots who have been subjected to rigorous medical and performance screening. The FAA defended its refusal to allow such a study of older pilots on the basis of the same argument that it uses to justify retention of the Age 60 Rule in general— namely that there is no reliable means for selecting pilots to participate in the study who were not at risk of sudden incapacitation or subtle deterioration in their abilities, and in the absence of such selection tests the study would represent too great a threat to aviation safety.
The flaws in the FAA’s reasoning discussed above make it difficult to credit the FAA’s conclusion that such a study would pose an unacceptable risk to aviation safety. If accommodating the reasonable expectation of commuter airlines and commuter pilots justifies delaying application of the Age 60 Rule to commuter operations, then surely the pressing need for data on older pilot health and performance justifies allowing a carefully screened group of pilots to fly past 60 as part of a systematic study of older pilots. And the FAA’s contention that there is no means by which to safely select participants for such a study is undercut by the agency’s claim that it can safely monitor younger pilots with *775known conditions that could result in sudden incapacitation or subtle deterioration in functioning. In addition, many experts who study the Age 60 Rule have argued that a carefully screened group of over 60 pilots could safely be allowed to fly as part of a systematic study. For example, the NIH report, which the FAA itself describes as “the most comprehensive study yet performed of the issues involved in age-related retirement of airline pilots,” Age 60 Rule, 60 Fed.Reg. at 65,982, recommended that the FAA undertake such a program and provided a general description of what would be required. Thus, clearly the authors of the NIH report, who were national experts on the medical conditions associated with aging, believed that such a study could be safely undertaken. See J.A. at 179-81; see also id. at 471-73 (testimony of Dr. James Hickman of the Mayo Clinic to the same effect); see generally Andreas E. Struck et ah, Multidimensional Risk Assessment versus Age as Criterion for Retirement of Airline Pilots, 40 J. Am. Geriatric Soc’y. 526, 530 (1992) (arguing that under an improved medical certification procedure conditions that might cause sudden incapacitation or subtle deterioration in functioning “would most likely be identified” and advocating an increase in the mandatory retirement age for pilots to 70), reprinted in J.A. at 622; Charles E. Drebing et al., Early Detection of Cognitive Decline in Higher Cognitively Functioning Older Adults, 8 Neuropsychology No. 1, 31, 35 (1994) (analyzing data on a battery of tests designed to screen for cognitive decline and concluding that the battery “exhibits a relatively high degree of accuracy”).
In any event, recent developments make it possible for the FAA to obtain medical and performance data from pilots flying in part 121 operations. Europe’s Joint Aviation Authority as well as several Asian countries now permit pilots to fly past age 60, and by treaty these pilots must be allowed to fly in U.S. airspace. As the majority notes, this fact creates a natural source of exactly the sort of information on older pilots that is needed. Another seemingly valuable source of data on pilot functioning are commuter pilots over 60 who will be allowed to fly for four years after commuter operations are shifted under part 121. A third potential source is non-part 121 operators who have agreed to allow pilots to fly past 60 pursuant to consent decrees with the Equal Employment Opportunity Commission. At no point in its decision, however, does the FAA suggest that it will make any efforts to obtain data on older pilot functioning from any of these sources.
Ordinarily, an agency’s failure to try and obtain data needed to initiate a rulemaking would not be arbitrary; agencies must be allowed to set their own agendas and allocate their resources as they see fit. Cf. Heckler v. Chaney, 470 U.S. 821, 831-32, 105 S.Ct. 1649, 1655-56, 84 L.Ed.2d 714 (1985). But circumstances surrounding the FAA’s retention of the Age 60 Rule are not ordinary. For nearly forty years, the FAA has argued that the Age 60 Rule is necessary to ensure the highest level of aviation safety but it has yet to supply the direct evidence of older pilot health and performance that is needed to support this proposition. See, e.g., Aman v. FAA, 856 F.2d 946, 948 (7th Cir.1988) (commenting that “the agency’s progress in developing an understanding of the relationship between aging and flight performance has been disappointing”). The Age 60 Rule also exacts substantial costs; pilots are forced to end their careers even though they are healthy and perfectly able to fly and the flying public is deprived of the valuable expertise that these pilots offer. These costs do not on their own make the Age 60 Rule unreasonable, since they may be unavoidable if aviation safety is to be assured only in this way, but their presence makes it incumbent on the FAA to try to obtain the evidence required to fully and accurately assess the need for the Age 60 Rule.
Most importantly, the Age 60 Rule stands as an instance of government-mandated age discrimination for a particular group of employees. The ADEA manifests our country’s rejection of measures that discriminate against individuals solely because of their age; its stated purpose was to “promote employment of older persons based on then-ability rather than age ... [and] to prohibit arbitrary age discrimination in employment.” 29 U.S.C. § 621(b). As I indicated earlier, I *776do not believe that the ADEA directly governs the FAA in its role as a regulator of aviation such as to make it necessary for the FAA to demonstrate that the Age 60 Rule is a bona-fide occupational qualification and therefore acceptable under the Act. But this does not mean that the FAA can ignore the ADEA altogether. The congressional condemnation of age discrimination embodied in the ADEA imposes a duty on the FAA to try to obtain data that might allow it to do away with its current reliance on an arbitrary across-the-board age cutoff as a method of ensuring aviation safety.6
III. Conclusion
Judges must be ever-vigilant to ensure that when enforcing the APA’s requirement of reasoned decisionmaking they defer to agency expertise. The importance of such deference is most acute in regard to safety determinations, given the potential catastrophic effects of inadequate safety regulations, and it is difficult to imagine an agency decision which judges would be more disposed to accept than one that implicates aviation safety. However, deference to agency expertise cannot be allowed to become toleration of arbitrary agency action'—or in this case inaction—even in an area as critical as aviation safety. Because I believe the FAA has failed to provide a reasoned explanation for its decision to retain the Age 60 Rule, I would remand to the agency for further proceedings.
. Up until 1995, part 121 governed common carrier operations involving aircraft with more than 30 seats or more than 7,500 payload capacity. In 1995, the FAA extended part 121 to cover airlines in scheduled passenger carrying operations with 10 to 30 seats and turbojets regardless of seating capacity. See Commuter Operations and General Certification and Operation Requirements, 60 Fed.Reg. 65,832 (1995).
. Although I disagree with the majority's analysis of the FAA's decision under the APA, I concur with the assessment that our usual standard of APA arbitrary and capricious review applies here, even though refusals to initiate rulemaking ordinarily are accorded particular deference by a reviewing court, because the FAA based its refusal to initiate a rulemaking to rescind the Age 60 Rule not on pragmatic resource concerns but on the merits of the rule. See Majority opinion ("Op.") at 763-64.
. In Baker, the Seventh Circuit commented that it was not clear "[e]xactly how this distinction [between younger pilots with known conditions and older pilots at risk of the conditions] applies as a practical matter/' but upheld the FAA because petitioners were seeking exemptions from the Age 60 Rule and thus bore the "burden [oí] present[ing] persuasive evidence that granting exemptions would not impair safety.” 917 F.2d at 322. Here, however, petitioners are challenging the rule itself under the APA, and the APA requires the FAA to provide a sufficient explanation demonstrating that its actions are reasonable. Cf. id. at 322 n. 6.
Contrary to the majority's thesis, Op. at 764-66, there is nothing in the record to support the idea that it is more difficult to monitor known medical conditions in an older pilot than in a younger pilot; nor is there anything—beyond the FAA's bare assertion that "there are no generally applicable medical tests that can, at this time, adequately determine which individual pilots are subject to” medical conditions associated with aging. Joint Appendix ("J.A.”) at 76—to support the notion that it is harder to diagnose such conditions in the first place in an older than in a younger pilot. The only possible difference is that disabling conditions—whether detected or not—may be more likely to occur in older than in younger pilots. But see infra note 4 (FAA's evidence is ambiguous on question of whether age 60 is the appropriate cut-off to guard against age-related decline). Yet without evidence that such conditions are more easily detected in younger pilots, this potential difference does not suffice to justify the FAA’s Age 60 Rule. Indeed, common sense suggests that examining physicians would be more likely to suspect, check for, and thus discover disabling conditions in the older pilots. Even if that were not the case, however, the possibility that older pilots are more subject to disabling medical conditions than younger pilots could be readily addressed by more frequent and thorough medical testing.
. In addition, if the possibility of an age-related cognitive decline is to be the linchpin of the FAA's defense of the Age 60 Rule, then the agency should provide a more in-depth discussion than it currently does of why 60 is an appropriate age cut-off to guard against this phenomenon. In its decision the FAA identifies the phenomenon of age-related cognitive decline, but provides no indication of when such decline impacts on an individual’s ability to function. The decision also comments that ''[o]ne in 10 persons over age 65 and nearly half of those over 85 have Alzheimer's disease alone, and increasingly it is found in people in their 40’s and 50's.” J.A. at 72. Without any indication of whether the incidence of Alzheimer’s increases with age before 65, this statement does not support an across-the-board age cutoff of 60. See also Institute of Medicine, Airline Pilot Age, Health and Performance: Scientific and Medical Considerations, J.A. at 215 (noting that the prevalence of dementia before 65 has been estimated at 1 percent, while the prevalence after 65 has been estimated at 5 percent); J.A. at 638-41 (affidavit of former Director of National Institute on Aging arguing that evidence on neurological and neuropsychological health does not support an age cut-off of 60).
. The majority also remarks that extension of the Age 60 Rule to air-taxi operations would not benefit petitioners, since it would only further restrict the opportunities available to pilots over the age of 60. But again, the petitioners’ challenge is to the FAA’s claim that the duty of air carriers to perform with the highest degree of safety is a sufficient basis to justify application of the rule to all part 121 operations, not to the FAA’s exclusion of air-taxi and other non-part 121 air carrier operations from the rule. If the FAA were to determine that the differences between different types of part 121 operations allowed pilots over the age of 60 to fly in some instances, more opportunities for older pilots would be created.
. A rider to the 1996 appropriations bill prohibited the National Transportation Safety Board from expending any funds to study the performance of pilots over 60. See Department of Transportation and Related Agencies Appropriations Act of 1997, Pub.L. No. 104-205, § 345, 110 Slat. 2951, 2976 (1996). But this rider was not in effect when the FAA rendered its decision, and thus is irrelevant to a determination of whether the FAA's failure to undertake measures to obtain data on older pilot functioning was reasonable.