Paccar, Inc., the builder of Kenworth and Peterbilt trucks, the American Trucking Associations, Inc. (ATA) and the Truck Equipment and Body Distributors Association (TEBDA) challenge Motor Vehicle Safety Standard No. 121 (the Standard) regulating air-braked trucks and tractor-trailers. 1 The Standard was issued by the National Highway Traffic Safety Administration (NHTSA) pursuant to the National Traffic and Motor Vehicle Safety Act of *635 1966 (the Act), 2 and is challenged here for the first time.

Congressional concern over the rising highway accident rate, the loss of lives and property incidental thereto, and the tendency of automobile manufacturers to concentrate on cosmetic design to the detriment of safety design, prompted passage of the Act. Congress voiced its belief that “the restrained and responsible exercise of Federal authority can channel the creative energies and vast technology of the automobile industry into a vigorous and competitive effort to improve the safety of vehicles.” S.Rep.No. 1301, 89th Cong., 2d Sess. 1, reprinted in [1966] U.S.Code Cong. & Admin. News pp. 2709, 2709.

To this end the Secretary of the Department of Transportation (Secretary) was given authority to regulate motor vehicle safety by the promulgation of appropriate minimum performance standards. 15 U.S.C. § 1392 . The purpose of the Act is to “reduce traffic accidents and deaths and injuries to persons resulting from traffic accidents.” 15 U.S.C. § 1381 . “[S]afety shall be the overriding consideration in the issuance of standards . . . .” 3 Each standard must meet the need for motor vehicle safety 4 and must be stated in objecfive terms. 15 U.S.C. § 1392 (a). The Secretary is instructed to consider “available motor vehicle safety data, including the results of research, development, testing and evaluation activities conducted pursuant to this chapter.” 15 U.S.C. § 1392 (f)(1). A separate section requires the Secretary to carry out research and development, which can be done by grants to “States, interstate agencies, and nonprofit institutions.” 15 U.S.C. § 1395 (a), (b). The Secretary is required to consider whether a standard is “reasonable, practicable 5 and appropriate for the particular type of motor vehicle” and “will contribute to carrying out the purposes of this chapter.” 15 U.S.C. § 1392 (f). He has authority to amend standards and to delay their effective date when the need arises. 6 The rulemaking power of the Secretary may be delegated to NHTSA. 49 C.F.R. § 501.2 .

Each manufacturer 7 of affected vehicles must certify that vehicles produced after the effective date of a standard conform to that standard. An exception is made for “incomplete vehicle manufacturers,” who may refuse to certify or provide a qualified certification. 8 With that qualification,' failure to certify, or false certification, is punishable by civil fines in amounts up to $800,- *636 000. 9 Moreover, any vehicle which is sold to a distributor or dealer, and thereafter found to be nonconforming, must be recalled by the manufacturer, at his expense, or brought into conformity where located. The manufacturer must pay the distributor or dealer “a reasonable reimbursement” at a rate that is fixed by the statute. 15 U.S.C. § 1400 (a).

Once NHTSA has indicated that a standard is final, any person “adversely affected” may petition the United States Court of Appeals for review. 15 U.S.C. § 1394 . The House Report 10 states that judicial review under the Act is based on the compliance provisions of the Food and Drug Act, 21 U.S.C. § 371 (f)(3), which contains a substantial evidence requirement. 11 The Act itself provides that the Administrative Procedure Act “shall apply to all orders establishing . a Federal motor vehicle safety standard,” 15 U.S.C. § 1392 (b), and the Administrative Procedure Act allows agency action to be held unlawful if “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706 (2)(A).

Whether there is a distinction between the tests of “substantial evidence” and “arbitrary and capricious” is the subject of vigorous debate, 12 but we see no pressing need to resolve that question here. The Secretary is required to file with the Court the evidence on which the standard was based. 15 U.S.C. § 1394 (a)(1). The factual basis of the Standard is subject to a “thorough, probing, in-depth review” and we are “required to decide whether the Secretary acted within the scope of his authority.” Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 415 , 91 S.Ct. 814, 823 , 28 L.Ed.2d 136 (1971). “Although, this inquiry into the facts is to be searching and careful, the ultimate standard of review is a narrow one. The court is not empowered to substitute its judgment for that of the agency.” Id. at 416 , 91 S.Ct. at 824 .

HISTORY OF THE STANDARD

In October, 1967, the predecessor of NHTSA announced its intention to promulgate a standard for air-braked trucks, tractor-trailers, and buses. 32 Fed.Reg. 14,278 (1967). A notice of proposed rulemaking was published in January, 1969, 34 Fed.Reg. 1055 (1969), and the proposed standard for trucks and buses was released on June 25, 1970, with a proposed effective date of January 1, 1972, 35 Fed.Reg. 10,368 (1970). 13 The proposed standard required vehicles to stop within certain distances, from certain speeds, without leaving a 12-foot wide lane, and without lockup of any wheel “more than momentarily.” It set forth brake actuation and release times, specified a relationship between brake chamber air pressure and brake retardation force (to be determined by placing each brake on an inertia dynamometer), provided for repetitious testing to assure brake recovery, and required a number of back-up systems for both system brakes and parking brakes. The initial stopping distance requirement from 60 mph was 217 feet on a dry (skid number 75) 14 surface.

The Standard was amended on numerous occasions, first, by increasing from 217 feet to 245 feet the distance in which a vehicle travelling at 60 mph was required to stop. This first increase was allowed “to more *637 accurately reflect the friction characteristics of a surface with a skid number of 75 . .” 15 The effective date of the Standard was extended until January 1, 1973.

In February, 1972, in response to petitions from manufacturers requesting reconsideration, the effective date was advanced to September 1, 1974, to “permit a longer period of fleet testing to evaluate the durability of the new systems . . .”37 Fed.Reg. 3907 (1972). NHTSA noted that “a number of petitions requested increases in the stopping distance required by S5.3.1. The distances specified are considered reasonable and well within the state of the art. Greater distances would increase the disparity between trucks and cars and be contrary to the interests of safety.” Id. These petitions were denied, without recitation as to the process by which the agency reached its determination.

The stopping distances prescribed by NHTSA were far shorter than the former stopping distances achieved by large trucks. Therefore, a completely new braking system was necessary 16 and, since the Standard also demanded that stops be made without lock-up of wheels (more than “momentarily”), most manufacturers notified NHTSA that an “antilock” device was required. 17 This device consists of a minicomputer attached to each axle of the truck or tractor-trailer which “senses” when a wheel begins to lock and momentarily releases the brake to regain traction. The agency indicated that the Standard did not “require” such a device but set forth criteria for it if used, 36 Fed.Reg. 3817 (1971), and noted its faith in the device’s reliability. 18

In May, 1974, NHTSA again extended the stopping distance, this time allowing 258 feet for a stop from 60 mph. The reason given was the necessity to account for “production variations.” 19 In addition, the effective date of the Standard was delayed until March 1, 1975, for trucks and buses, and until September, 1975, for trailers. 39 Fed.Reg. 20,350 (1974).

In early 1974, the agency began to receive continual complaints to the effect that antilock devices were unreliable, and rendered vehicles dangerously unsafe when combined with new, more effective brakes. 20 In 1974, NHTSA replied that the *638 Standard was practicable as written, since its own investigation of the major antilock manufacturers indicated “millions of axle miles of antilock operation with a malfunction rate comparable to other equipment presently in highway service, and no highway accidents attributable to the device.” The agency announced that the Standard was “final” for purposes of review. 39 Fed.Reg. 39,881 (1974).

But “finality” did not mean an end to amendments, nor to industry protestations. 21 NHTSA acknowledged that production variables were affecting performance, 22 and moved to amend the Standard to increase the stopping distance to 277 feet at 60 mph. 40 Fed.Reg. 38,160 (1975). This amendment was calculated to allow manufacturers to “depower” the front axles, relieving the danger of lockup in the event of antiloek failure. The agency conceded that the requirements of the original Standard were difficult to achieve in a mass production situation. 23 In a continuing attempt to find a reasonable stopping distance within the technological reach of a mass production industry, the agency announced that a public meeting was necessary to explore the complaints about the safety of the antilock systems, and to “fully consolidate and evaluate such data.” 24 This was not the first public meeting called by NHTSA, but it *639 was the first to concentrate on the antiloek system, the stopping distance requirements, and the dangers. In October, 1975, nine months after the Standard went into effect for trucks and buses, a public hearing was held in Washington, D. C., with members of the industry and public invited to attend to comment “on the record.” 25 As a result, NHTSA proposed and effected a one-year suspension of the Standard for buses, because of “a pattern of erratic behavior of bus antilock equipment.” 40 Fed.Reg. 52,-856 (1975).

The agency also found that “some vehicles produced to comply with the standard demonstrate unsatisfactory handling characteristics. The source of the problem is not technological unfeasibility of producing vehicles with desirable handling characteristics, as demonstrated by many of the 121-equipped vehicles produced since March 1, 1975.” 41 Fed.Reg. 8786 (1976). NHTSA found that large compliance margins and component variability were the cause of the manifest problems. While indicating that it could require redesign to solve these problems, NHTSA chose instead to extend the stopping distances once more, allowing a vehicle travelling 60 mph to stop within 293 feet. Id. Additional extensive changes of the Standard as it affects trucks and tractor-trailers, 40 Fed.Reg. 56,920 (1975), were proposed. At the time of oral argument in our court, three amendments were still pending. 41 Fed.Reg. 29,696 (1976). The effective date of the Standard as it applies to buses was postponed until January, 1977, because of severe antilock problems. 40 Fed.Reg. 52,856 (1975).

THE MERITS

Paccar and ATA have concentrated their attack on the Standard’s stopping distance criteria, i. e., the antilock requirement. 26 All petitioners argue that the testing procedures promulgated by NHTSA are “nonobjective” and TEBDA individually objects to the certification requirements of the Standard.

Paccar and ATA vigorously assert that the Standard does not meet the statutory requirement of “practicability” because no need has been demonstrated for this air brake system, the cost of the system is high, 27 and its requirements are beyond the reach of current technology. They argue that it is not a “reasonable” standard, since the failure of what they consider an unreliable device, the antilock, in a panic braking situation creates a potentially more dangerous situation than a “prestandard” vehicle because of the new, stronger braking system.

*640 In charging that no “need” has been demonstrated for this Standard, Paccar and ATA ignore the extensive record compiled by the agency. Congress was aware that “it is virtually impossible to obtain specific information and data concerning the causes of traffic accidents . . . .” H.R.Rep. No.1776, supra note 5, at 11. To combat this uncertainty, the Secretary was empowered to implement standards and conduct research. Two studies commissioned by NHTSA found that improving truck and tractor-trailer stability, by preventing jackknifing and wheel lockup, would substantially increase the safety of these massive vehicles on the highway. 28 Several studies cited in the administrative record indicate that greater braking capability would improve highway safety. 29 In addition, the evidence produced by NHTSA supports the proposition that in down-hill accidents, trucks (undifferentiated from tractor-trailers) are the striking vehicle far more frequently than their numbers on the road would dictate. 30 There is evidence that the size and weight of these vehicles can produce disastrous results to occupants of automobiles when an accident occurs 31 and that the jackknifing of tractor-trailers can produce “chain-reaction” accidents. 32 While there is no direct evidence that shorter braking distances would prevent a particular number of accidents, the conclusion from one study was that trucks travelling at an “unreasonable speed for conditions, not necessarily above the posted speed limit” were involved in nearly 43% of the accidents investigated. 33 Common sense would seem to indicate that when two types of vehicles of widely divergent size share the same roadway, and the larger, heavier vehicle has a much longer stopping distance, any situation which reqúires emergency braking is potentially disastrous to the smaller automobile and its passengers.

It has been asserted that improved brake maintenance would eliminate accidents attributable to braking systems without the need to install new brake mechanisms. While there is some evidence that better brake maintenance would solve a number of problems, 34 we are satisfied, from our review of the data, that NHTSA was justified in promulgating a Standard requiring improved air brake systems and stability mechanisms for the industry.

The goals of the Standard, decreasing the disparity of stopping distances between ordinary automobiles and heavier vehicles, 37 Fed.Reg. 3905 (1972), and maintaining vehicle stability and directional control, 41 Fed. Reg. 8784 (1976), are both reasonable and practicable. However, after a careful review of the extensive record in this case, we have come to the conclusion that because of unforeseen problems in the development of the new braking systems, the Standard was neither reasonable nor practicable at the time it was put into effect.

*641 The House Report stated: “Standards, of course, cannot be set in a vacuum. They must be based on reliable information and research.” H.R.Rep.No.1776, supra note 5, at 11. To this end, the agency was commanded to conduct such research itself, or to secure the results of research by contracts with “States, interstate agencies, and nonprofit institutions.” 15 U.S.C. § 1395 (a), (b). We are not persuaded that the Standard, as finally issued, was based on the quality of “reliable information and research” that the Act required as a prerequisite.

NHTSA originally proposed a standard that it considered well within the “state of the art.” Antilock devices had been used on some cars and aircraft. 35 Some trucks were easily able to meet the stopping requirements of the Standard. 41 Fed.Reg. 8785 (1976). But the critical problems began with mass production of vehicles designed to meet the Standard.

The Standard required redesigning the entire braking systems of many trucks and tractor-trailers by increasing front brake torque and strengthening the brake design to enable the vehicles to meet the stopping distance requirements. 39 Fed.Reg. 7969 (1974). One manufacturer informed NHTSA that it was required to introduce 40,000 new part numbers in its system alone as the result of the Standard. 36 As new components — strong front axles, new brake linings, and mini-computers to control wheel lock and skid — were mass-produced, and subsequently put together in a unique combination, grave problems developed. There were complaints about electrical problems, relay valve failures, false warning signals, and failures of the warning signal 37 required by NHTSA when antilock is used. 38 The agency was soon informed of a number of situations in which the warning light did not flash when the system was inoperable 39 and of other cases in which the warning light flashed continuously. The driver or fleet manager sometimes “solved” the latter problem simply by disconnecting the signal. 40 And we are informed that Greyhound Lines, anticipating the inevitable danger to its passengers from activating the antilock, simply disconnected the device when its newly equipped vehicles were delivered. Radio frequency interference (also called magnetic frequency interference) was an unanticipated problem 41 causing the antilock to malfunction and the brakes to fail without warning. 42 This problem was *642 still being encountered a year after the Standard had been made effective. When the antilock system failed, the high torque front brakes, in the opinion of the industry, created a serious safety problem. 43 Among the problems encountered were lateral instability, jackknifing, wheel lockup, and “pulling” to one side. 44

Although the Act requires manufacturers to report all safety-related defects to NHTSA, 15 U.S.C. § 1411 , manufacturers informed the agency that truck users did not always advise them of the problems. Consolidated Freightway representatives testified that a spot check of their Standard 121-equipped trailers disclosed that 65% of the antilock devices malfunctioned. 45 General Motors reported that, despite instructions to their users to report defects, a spot check by the company revealed a 59% nonoperative rate for the antilock system. 46 In March, 1976, another survey, also by Consolidated Freightway at the request of the Subcommittee on Consumer Protection and Finance of the House Interstate and Foreign Commerce Committee, discovered that 48% of the trailers had malfunctioning anti-locks. 47 The absolute number of vehicles represented by these figures is quite small. But in view of the testimony to NHTSA that drivers were disconnecting warning lights, that large numbers of antilock systems certified at the factory were malfunctioning upon delivery or within a short time after delivery, 48 and that malfunctions created a hazard, we think that the agency had a responsibility to examine the results of its rulemaking by investigating more fully the safety of vehicles in use.

Although NHTSA received requests to postpone the March, 1975, effective date, no further delay in the truck and tractor standard was allowed by the agency. 49 While we can understand the agency’s reluctance again to delay the effective date, after having postponed it several times previously, 50 we agree with the Sixth Circuit that NHTSA has the responsibility to assure that the new systems it requires are reliable when placed in use. Chrysler Corp. v. Department of Transp., 472 F.2d 659 (6th Cir. 1972). If, to accomplish that end, the industry demonstrates that it must be given more time to solve its start-up problems, the agency must revise its timetables. As the court stated in Chrysler Corp.:

Id. at 673.

The Act requires that standards set by the agency protect the public against unreasonable risk of accidents “occurring as a *643 result of the design, construction or performance of motor vehicles.” 15 U.S.C. § 1391 (1) (emphasis added). NHTSA has the power to determine whether performance conforms to theory through its mandate to research or contract for research, to do compliance testing, and to “conduct any inspection or investigation . . . which may be necessary to enforce” federal vehicle safety standards. 15 U.S.C. § 1401 (a)(1). Yet, despite massive evidence presented to NHTSA that many vehicles equipped with antilocks were potentially more dangerous in performance than pre-Standard vehicles, the agency admitted during oral argument in July, 1976, that it had done no compliance testing since the Standard went into effect over a year earlier. Agency attorneys also admitted that prior to the effective date of the Standard, NHTSA had done no reliability testing of vehicles with Standard-121 components whatsoever, other than the testing reviewed in the Murphy Report, 51 which did not include stops at the higher speeds. For “proof” of the reliability of the system, the agency has relied only upon the testimony of some manufacturers, 52 the lack of reports of highway accidents “caused” by antilock, 53 and the results of testing a mere three 121-equipped tractor-trailers from June until September, 1975. But because antilock problems have appeared with usage rather than at the manufacturing stage, the foregoing methods employed by NHTSA to examine the validity of complaints were inadequate. 54 The agency has a heavy responsibility indeed, a responsibility to ascertain, with all reasonable probability, that its regulations do not produce a more dangerous highway environment than that which existed prior to governmental intervention. In view of substantial evidence received by NHTSA that the “in-use” performance of vehicles equipped with antilock was not consistént with the performance required by the Standard, and in some circumstances may have been more hazardous than the performance of pre-Standard vehicles, the agency’s failure to conduct more intensive testing of trucks and tractor-trailers certified under the Standard was not only unreasonable, but, also, a legal abuse of its discretion.

There is much in the Standard that has long been needed for highway safety, and it is undisputed that the antilock device, when perfected, will advance that goal. We are also aware that some manufacturers have expended a great deal of time, energy, and money over the last six years, in attempting to produce vehicles to comply, and that they have, to a large degree, been successful. However, the comprehensive data before us indicate that there is a strong probability that the performance of a substantial number of vehicles whose braking systems have been altered to include “overpowered” front axles and antilock devices has created a potentially more hazardous highway situation than existed before the Standard became operative. Therefore, until such time as NHTSA develops evidence that the new braking systems required by the Standard do not create the possibility of greater danger to the public, those parts of the Standard requiring heavier axles and the antiloek device should be suspended. The evidence indicates that this can be accomplished if we hold, as we do, that the stopping distance requirements from 60 mph are invalid. The agency may, of course, continue to propose new standards, including the promulgation of new stopping distance requirements to replace those here invalidated. We hold only that more provative and convincing data evidencing the reliability and safety of vehicles that are equipped with antilock and in use must be available before the agency can enforce a standard requiring its installation.

*644 TESTING REQUIREMENTS

To ensure compliance with the Standard, NHTSA has retained the right to test certified 121 vehicles on testing tracts with road surface skid numbers of 75 and 30. See 49 C.F.R. § 571.121 . The skid number represents the coefficient of friction between the tire and the road surface. It is a quantitative measure of the slickness of the road: skid number 75 represents a dry road, skid number 30 a wet road. Manufacturers whose vehicles fail NHTSA’s tests and who cannot establish that their own testing procedures reflected due care are subject to administrative penalties. Furthermore, nonconforming vehicles are subject to the recall and repair provisions of 15 U.S.C. § 1400 (a). Petitioners raise several objections to NHTSA’s testing procedures.

Petitioners argue that the skid numbers of 75 and 30 are ill-chosen. We agree. The skid numbers are based on a mathematical calculation that assumes the use of a particular tire. It is uncontroverted that the particular tire is no longer produced and that the use of available tires produces different results. 55 It is also uncontroverted that there was a mathematical error in the original formula for determining the skid numbers. 56

Petitioners also raise a more basic objection to the skid number method of testing stopping distances. They assert that the method is neither “objective” nor “practicable,” as required by 15 U.S.C. § 1392 (a), because the skid number of a given .surface varies over time and with use. 57

We cannot agree that the test is not “objective.” The fact that it is difficult to create and thereafter maintain a road surface with a particular coefficient of friction does not render the specified coefficient any less objective. The Sixth Circuit, in examining an attack on the same testing regulations, defined an objective standard as one whose testing procedures are “capable of producing identical results when test conditions are exactly duplicated, that they be decisively demonstrable by performing a rational test procedure, and that compliance is based upon readings obtained from measuring instruments as opposed to the subjective opinions of human beings.” Chrysler Corp. v. Department of Transp., 472 F.2d 659 , 676 (6th Cir. 1972) (footnote omitted). The skid number method of testing braking capacity meets this definition. Identical results will ensue when test conditions are exactly duplicated. The procedure is rational and decisively demonstrable. Compliance is based on objective measures of stopping distances rather than on the subjective opinions of human beings.

We do agree with petitioners, however, that NHTSA’s testing procedure is not “practicable.” The fluctuations in skid numbers on a given road surface make it impracticable for manufacturers to conduct tests that will assure that their vehicles will exactly meet the objective standard when tested by NHTSA. Their only recourse will be, not simply to comply with the stated standard, but to overcompensate by testing their vehicles on road surfaces substantially slicker than official regulations require.

NHTSA responds that it has met this objection by testing compliance on a road surface substantially more sticky (skid number 85) than required by the Standard (skid number 75), thus allowing a margin sufficient to compensate for any variance in the skid number road surfaces used by manufacturers. These informal assurances are not enough. Manufacturers are entitled to testing criteria that they can rely upon with certainty.

In the same vein, petitioners complain that NHTSA has not specified the time intervals between tests, the duration of permissible wheel lockup during tests, and the amount of curving in testing track roadways. Again NHTSA responds with informal assurances that it will act reasonably in *645 performing its tests. In these respects also the manufacturers are entitled to formal and reasonably specific testing criteria.

Petitioners’ argument that NHTSA should promulgate a regulation formalizing its informal agreement to test vehicles with a load that is within the “center of gravity envelope” of the vehicle 58 has been mooted during the pendency of these proceedings. In May of 1975 NHTSA promulgated just such a regulation. 40 Fed.Reg. 21,032 (1975).

TEBDA’s challenge to the certification requirements of the Standard is also moot. TEBDA represents approximately half of the “intermediate” and “final stage” manufacturers in this country. The “first stage” or “incomplete” manufacturers build the truck chassis, and TEBDA members modify that chassis to produce dump trucks, cement trucks, garbage trucks, and other trucks. TEBDA complained that chassis-cab manufacturers were not required to certify their work and that, as a result, TEBDA members were forced to certify work over which they have had no control. During the pendency of these proceedings NHTSA modified its regulations to require chassis-cab manufacturers to certify that their vehicles will conform to the standard if completed in accordance with specified instructions. See 42 Fed.Reg. 37,814-18 (1977) (to be codified in 49 C.F.R. §§ 567 , 568); 40 Fed.Reg. 45,847-49 (1975).

TEBDA also challenges as not “practicable” the road testing procedure set out at great length at 49 C.F.R. § 571.121 . It is undisputed that it would be economically unfeasible for TEBDA members to road test each vehicle they complete. 59 TEBDA argues, however, that its manufacturers run the risk of violating the “due care” provision of 15 U.S.C. § 1397 (a)(1)(C) if they do not road test, since road testing is the only method NHTSA has specified whereby manufacturers can ensure that their vehicles are properly certified.

NHTSA asserts that road testing is not required of the manufacturers, and it is only necessary that any truck manufactured meet the test that it performs. According to the agency, due care could be satisfied by any number of alternative methods, such as group testing, mathematical calculations, and so forth. 60 TEBDA contends that, since such alternatives are not contained in the regulations, its manufacturers have no assurance that these alternatives will ultimately be found to comply with the due care requirement.

In our opinion, the plain statutory mandate of “practicability” and “objectivity” is not met by agency “suggestions” of what might constitute compliance with the amorphous due care standard. Since NHTSA has admitted that road testing is beyond the practical and financial reach of the final stage manufacturers, it must propose some alternative method for those manufacturers which, if followed, it will recognize as fulfilling the due care requirement.

Our decision in no way invalidates the testing procedures, nor does it affect the requirement that each vehicle sold to a distributor or dealer must conform to a legally appropriate standard. 15 U.S.C. § 1400 . We hold only that this regulation is neither “objective” nor “practicable” because it could subject intermediate and final stage manufacturers to the penalties of 15 U.S.C. § 1397 simply because they could not perform the only procedure specified in the regulations for fulfilling the due care provision. Statutes prescribing penalties, civil or criminal, must be drafted without ambiguity. Successive authorities of NHTSA might take an entirely different view than that announced by the incumbents, and sub *646 jeeting the manufacturers to such a risk does not comport with due process requirements.

We have concluded that numerous other issues raised by petitioners are without merit.

The cause is remanded to the Respondent for further proceedings not inconsistent with the views herein expressed.

SO ORDERED.

. 49 C.F.R. § 571.121 . The part of the Standard that regulates air-braked buses is not an issue here.

. 15 U.S.C. § 1381-1431 .

. S.Rep.No. 1301, 89th Cong., 2d Sess. 6, reprinted in [1966] U.S.Code Cong. & Admin. News pp. 2709, 2714.

. “Motor vehicle safety” is defined in the Act as:

15 U.S.C. § 1391 (1).

. “Practicable” is defined to “require consideration of all relevant factors, including technological ability to achieve the goal of a particular standard as well as consideration of economic factors.” H.R.Rep.No. 1776, 89th Cong., 2d Sess. 16 (1966).

. Congress foresaw that such postponements might be necessary “because it may be a practical economic and engineering impossibility, as well as a source of great hardship and unnecessary additional cost, to require that all vehicle changes required by any new safety standard ... be accomplished . . . within 1 year.” S.Rep.No. 1301, supra note 3, at 7, reprinted in [1966] U.S.Code Cong. & Admin.News pp. 2709, 2714.

. “ ‘Manufacturer’ means any person engaged in the manufacturing or assembling of motor vehicles or motor vehicle equipment, including any person importing motor vehicles or motor vehicle equipment for resale.” 15 U.S.C. § 1391 (5).

. 49 C.F.R. § 568.4 (a)(7)(iii). An incomplete manufacturer that builds a chassis-cab, which is thereafter turned into a specialized truck by a “final stage manufacturer,” must certify that the completed vehicle will conform to the standard if specified instructions are followed. See text accompanying note 58 infra.

. 15 U.S.C. §§ 1397 (a)(1)(C), 1398. When a vehicle is certified falsely, the provisions of these sections apply only if the person certifying “in the exercise of due care has reason to know that such certificate is false or misleading in a material respect.” 15 U.S.C. § 1397 (a)(1)(C).

. H.R.Rep.No. 1776, supra note 5, at 16.

. See also S.Rep.No. 1301, supra note 3, at 8, reprinted in [1966] U.S.Code Cong. & Admin. News pp. 2709, 2715-16.

. See K. Davis, Administrative Law of the Seventies §§ 29.00 to .01-5 (1976).

. The first proposal regulating air brakes on trailers was published at 35 Fed.Reg. 10,456 (1970) with the same proposed effective date of January 1, 1972.

. The “skid number” roughly measures the friction between the tire and the pavement: the “stickier” the pavement the higher the skid number.

. 36 Fed.Reg. 3817 (1971). See also 37 Fed. Reg. 12,496 (1972).

. As described by Rockwell: “Front axles require redesign to carry the added combined load. The lower rated axles require redesign in all components, the higher rated ones require minimum redesign. All of the mountings for the brakes must be strengthened.” Letter from Rockwell to NHTSA (Oct. 12, 1971) (Doc. No. 70-16-PRC-036). See also Letter from Ford Motor Co. to NHTSA (Jan. 14, 1974) (Doc. No. 121-PRM-000031); Letter from General Motors Corp. to NHTSA (Nov. 28, 1973) (Doc. No. 70-17-No4-001).

. Manufacturers have indicated that, to meet this requirement, they will supply substantial braking torque on front axles and will prevent uncontrolled wheel lock-up in most cases by use of newly-developed antilock control systems. Most requests to modify the standard are prompted by uncertainty as to the availability and reliability of the necessary components involved in these changes.

39 Fed.Reg. 7966 (1974).

. The NHTSA has found that antilock systems are reliable and offer significant improvement in vehicle braking characteristics.

39 Fed.Reg. 7967 (1974).

. Most manufacturers argued that the 5 percent longer distances would be required for the indefinite future, because production variations would continue to affect performance significantly. The NHTSA established the stopping distances on the basis of the ability of available equipment, and expects that experience in the production of these components will lead to predictable quality and the assurance that a vehicle will in fact perform as well as it is designed to.

39 Fed.Reg. 17,553 (1974).

. “[W]e believe that a most serious safety problem may be created by present state-of-the-art brake systems designed to meet MVSS121. This is due to the possibility of complete loss of vehicle control resulting from a failure of the wheel lock control system.” Letter from *638 General Motors Corp., supra note 16; see 39 Fed.Reg. 17,550 (1974); Letter from Ford Motor Co. to NHTSA (Apr. 1, 1974) (Doc. No. 74-10-No 1-023).

. See 41 Fed.Reg. 8783 (1976); 41 Fed.Reg. 1598 (1976); 40 Fed.Reg. 38,160 (1975); 40 Fed.Reg. 21,031 (1975); 40 Fed.Reg. 12,797 (1975); 40 Fed.Reg. 8953 (1975). ATA wrote the agency:

Letter from ATA to NHTSA (Mar. 25, 1975) (Doc. No. 74-10-No15-004).

. General Motors . . indicated that similar 121 vehicles can register as much as a 20-percent difference in stopping distances as a result of uncontrolled variability in brake component performance. International Harvester, which until recently had supported 5-percent longer stopping distances on an interim basis, now points to certain variables, including brake linings, in requesting long distances on a permanent basis. Diamond Reo reported the same experience .

40 Fed.Reg. 12,797 (1975).

40 Fed.Reg. 24,915 (1975).

. [T]his agency has evaluated the submitted data which show variability of various components and concludes that the degree of variability can necessitate more aggressive brake packages than are desirable to achieve the stopping distances contemplated in development of the standard. One undesirable characteristic cited by PACCAR is the side-to-side handling problems which can arise during severe braking associated with any imbalance of the brakes. Ford data demonstrate the point at which tire aligning torque becomes negative and contributes to steering instability under severe braking. Other manufacturers conclude that the variability in brake assembly and other areas forces use of brake packages which approach the maximum in technological feasibility and leave no room for a “compliance margin.” In agreement that these problems can arise in some cases, NHTSA proposes an increase in the stopping distance permissible in a stop from 60 mph in the loaded condition on a test surface with a skid number of 75 (with corresponding changes at lower speeds).

40 Fed.Reg. 24,916 (1975).

. The NHTSA has received reports that the new-introduced (sic) braking systems and antilock systems of some manufacturers, have not been operating properly, causing increased maintenance and the possibility of unsafe vehicle braking performance. Eight safety-related defects in the production sys *639 terns have been reported to the NHTSA, and recall campaigns are underway at this time. ******

To ensure that the data upon which the NHTSA relies are complete, the agency considers it imperative to collect all existing information on the operation of the new systems as soon as possible. For these reasons the NHTSA has determined that ... a public meeting is desirable to discuss the field experience accumulated with the new vehicles.

40 Fed.Reg. 43,050 (1975).

. Unlike most prior public meetings, the October meeting was recorded and transcribed.

. Although NHTSA denies that antilock is required by the Standard, extensive testimony and evidence from the industry, unrebutted agency evidence, convinces us, as a practical matter and beyond doubt, that antilock was necessary to meet the Standard. See, e. g., 39 Fed.Reg. 7966 (1974); Letter from Ford Motor Co., supra note 16; Letter from General Motors Corp., supra note 16; Letter from International Harvester Co. to NHTSA (Mar. 26, 1971) (Doc. No. 70-16-PRC-001); Letter from Wagner Electric Corp. to NHTSA (Apr. 10, 1975) (Doc. No. 74-10-No15-010); Testimony of Consolidated Freightways, Inc. at TR-15 (Oct. 1975) (Doc. No. 75-16-No2-042).

. At the public hearings truck manufacturers testified that the following costs are attributable to the Standard: Ford Motor Co., $23 million, Testimony at TR-351 (Oct. 1975) (Doc. No. 75-16-No2-042); General Motors Corp., $15 million, Testimony at TR-360; Kelsey-Hayes, $25 million, Testimony at TR-388. The antilock system adds $1200 to $1500 to the cost of each truck. Testimony of National Ass’n of Motor Bus Owners at TR-63; Testimony of Greyhound Lines, Inc. at TR-70, 88; Testimony of Transport of New Jersey at TR-128; Testimony of Leaseway Transp. Corp. at TR-228. The President’s Council on Wage and Price Stability estimated that the Standard would cost as much as $400 million per year in capital costs alone. Testimony at TR-306-07.

. R. Murphy, R. Limpert & L. Segal, Highway Safety Research Institute of the University of Michigan, Bus, Truck, Tractor-Trailer Braking System Performance (Mar. 1971) [hereinafter cited as Murphy Report]; National Transportation Safety Board, Special Study — Commercial Vehicle Braking (Nov. 1972).

. Federal Highway Administration, U.S. Dep’t of Transportation, Emergency Application Systems for Power Brake Mechanisms on Highway Trailer Combinations (Aug. 1970); Murphy Report, supra note 28.

. Scott & O’Day, Statistical Analysis of Truck Accident Involvements (1971) (Contract HS-800-627); Baker, Wong & Masemore, Fatal Tractor-Trailer Crashes: Considerations in Setting Relevant Standards (July 14, 1975) (Doc. No. 75-16-No1-067).

. NHTSA, Update of Information on Heavy Truck Accidents (Dec. 8, 1975) (Doc. No. 75-16-No5-056); Scott & O’Day, supra note 30; R. Wolf, Cornell Aeronautical Laboratory, Truck Accidents and Traffic Safety — An Overview (1968).

. Baker, Wong & Masemore, supra note 30.

. Bureau of Motor Carrier Safety, U.S. Dep’t of Transportation, Summary of Accident Investigations (1969). See also Baker, Wong & Masemore, supra note 30.

. Brakes, Fleet Owner, Feb. 1963, at title page; R. Wolf, supra note 31; Testimony of Captain E. Kynaston, California Highway Patrol, at TR — 485 (Oct. 1975) (Doc. No. 75-16-No2-042).

. 30 Fed.Reg. 17,552 (1974); NHTSA, A Procedure for Evaluating Vehicle Braking Performance (1971) (Contract HS-31-051); Testimony of G. Eads, Council on Wage and Price Stability, at TR-318 (Oct. 1975) (Doc. No. 75-16-No2-042).

. Letter from International Harvester Co. to NHTSA (1975) (Doc. No. 75-16-No2-005).

. NHTSA, Listing of Safety Defect Recall Campaigns (1975) (Doc. No. 75-16-No2-010); Letter from ATA to NHTSA (Mar. 29, 1974) (Doc. No. 74-10-Nol-014); Letter from NHTSA to Weyerhauser Co. (Mar. 8, 1976) (Doc. No. 75-16-No5-067); Testimony of Holly Farms Poultry Indus. at TR-176-78 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of Wesco Truck & Trailer Sales at TR-518-19.

. 41 Fed.Reg. 8786 (1976); 30 Fed.Reg. 7967 (1974).

. NHTSA, Listing of Safety Defect Recall Campaigns, supra note 37; Letter from Hy-Vee Food Stores to NHTSA (Mar. 1976) (Doc. No. 75-16-No2-173); Testimony of Leaseway Transp. Corp. at TR-231-33 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of Roadway Express at TR-269; Testimony of Freightliner Corp. at TR-296.

. Petition for Reconsideration, International Harvester Co. (Mar. 22, 1972) (Doc. No. N3 70-16-1).

. 41 Fed.Reg. 8787 (1976); 40 Fed.Reg. 59,228 (1975); Testimony of Rockwell Int’l Corp. at TR-150 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of General Motors Corp. at TR-361.

. NHTSA, Memo of Visit to International Harvester Co. (Mar. 12, 1976) (Doc. No. 75-16-No26-002); Report from Kelsey-Hayes Co. to NHTSA (Jan. 31, 1975) (Doc. No. 75E-031); Letter from NHTSA to Weyerhauser Co., supra note 37; Testimony of Freightliner Corp. at TR-296 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of General Motors Corp. at TR-361; Testimony of Kelsey-Hayes Co. at TR-391; Testimony of California Highway Patrol at TR-466. Citizens Band radio transmission is not a problem. Letter from NHTSA to Weyerhauser Co., supra note 37; Letter from NHTSA to Congressman Shuster (Sept. 17, 1975) (Doc. No. 75-16-NO2-032).

. Letter from ATA to NHTSA (Mar. 25, 1975) (Doc. No. 74-10-No5-065); Letter from Ford Motor Co., supra note 16; Letter from General Motors Corp. to NHTSA (Apr. 1, 1974) (Doc. No. 74-10-No1-26); Letter from General Motors Corp., supra note 16; Testimony of Wilson Freight Co. at TR-211 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of Leaseway Transp. Corp. at TR-234; Testimony of Queens Transit Corp. and Steinway Transit Corp. at TR-510.

. NHTSA, Internal Progress Reports B-7 (1976) (Doc. No. 75-16-No2-183).

. Testimony at TR-17-24 (Oct. 1975) (Doc. No. 75-16-NO2-042).

. Letter from General Motors Corp. (Apr. 1, 1974), supra note 43.

. Letter from Consolidated Freightways, Inc. to the Secretary (Apr. 6, 1976) (NHTSA Reliability File No. 20).

. Letter from A-P-A Transport Corp. to NHTSA (Feb. 5, 1976) (Doc. No. 75-16-No2-010); Letter from J. P’ Silvey Trucking to NHTSA (Oct. 7, 1975) (Doc. No. 75-16-No2-052); Testimony of Leaseway Transp. Corp. at TR-230 (Oct. 1975) (Doc. No. 75-16-No2-042); Testimony of Roadway Express at TR-276.

. 40 Fed.Reg. 1248 (1975); 39 Fed.Reg. 40,168 (1974); 39 Fed.Reg. 39,880 (1974).

. 40 Fed.Reg. 1246 (1975); 39 Fed.Reg. 28,161 (1974); 30 Fed.Reg. 20,380 (1974); 39 Fed.Reg. 17,550 (1974); 37 Fed.Reg. 3905 (1972).

. Murphy Report, supra note 28.

. 41 Fed.Reg. 8785 (1976); 40 Fed.Reg. 5822 (1975); 39 Fed.Reg. 39,880 (1974); 39 Fed.Reg. 17,552 (1974).

. 39 Fed.Reg. 39,880 (1974).

. The agency’s evidence does not even show whether the above-mentioned antilock defects occur only in certain models of the device or only when placed on a vehicle of a certain type or weight. This is one of the many gaps in NHTSA’s proof that it must undertake to fill.

. NHTSA, Braking Systems (Apr. 1975) (Doc. No. 75-27-GR-001).

. Id.

. Letter from Mack Truck to NHTSA (July 1975) (Doc. No. 75-16-No1-023).

. Letter from NHTSA to Trailmobile Corp. (May 8, 1976) (Doc. No. 75-16-No7-007).

. There are only four test tracks in the United States, and the cost of testing each vehicle is between $2000 and $4000. Letter from NHTSA to the Honorable Warren G. Magnuson (May 1975).

. Letter from NHTSA to TEBDA (Apr. 16, 1976); Letter from NHTSA to TEBDA (Mar. 6, 1975) (denial of Petition for Rehearing); NHTSA References.