Opinion for the Court filed by Senior Circuit Judge MacKINNON.
MacKINNON, Senior Circuit Judge:
Title II of the Clean Air Act (the Act), 42 U.S.C. §§ 7521-7574 (Supp. V 1981), deals with air pollution caused by “moving sources,” i.e., motor vehicles. In the 1970 Amendments to the Clean Air Act,1 Congress determined that emission inspections of vehicles while in use were essential to controlling air pollution. To this end, vehicle manufacturers were required to warrant compliance with emission standards throughout the useful life of the vehicles they produced. Section 207(b) of the Act, 42 U.S.C. § 7541(b), directs the Administra[242]*242tor of the Environmental Protection Agency (EPA) to establish, for use in state and local inspection and maintenance programs, “short tests” which could reasonably identify vehicles that exceeded emission standards.
Petitioners2 seek judicial review of the short test regulations promulgated by the Administrator.3 Petitioners contend that the Administrator’s action was arbitrary and capricious and that the regulations exceed statutory authority. Because we deem the short test regulations to be a reasonable exercise of the Administrator’s discretion under the Act, we affirm the regulations in their entirety.
I.The Statutory Scheme
In 1970, Congress enacted in Title II of the Act a powerful and comprehensive strategy for controlling pollution from mobile sources. Section 202 of the Act, 42 U.S.C. § 7521, prescribes standards for emissions of hydrocarbon (HC), carbon monoxide (CO), and oxides of nitrogen (NOx) for certain classes of motor vehicles. These standards must be met during the “useful life” of the vehicle; for passenger cars, the Act defines “useful life” as five years or 50,000 miles, whichever occurs first.4 Id. § 7521(d).
Section 206 of the Act, 42 U.S.C. § 7525, directs the Administrator to require the testing of new vehicles to determine whether they conform to emission standards. A manufacturer may not sell a new vehicle or engine that has not been issued a certificate of conformity. EPA has already established “certification” procedures to test preproduction, prototype vehicles. The procedures, which are not at issue in this case, are known as the Federal Test Procedure (“Federal Test”). The Federal Test is a timeconsuming and highly complex procedure for simulating driving situations under closely controlled laboratory conditions. See Emission Regulations for 1977 and Later Model New Light-Duty Vehicles and New Light-Duty Trucks; Test Procedures, 40 C.F.R. pt. 86, subpt. B. (1982).
Acknowledging that presale testing alone could not guarantee that vehicles would comply with emission standards throughout their useful lives, Congress also established two warranty schemes under section 207 of the Act, 42 U.S.C. § 7541. See H.R.Rep. No. 91-1783, 91st Cong., 2d Sess. 51 (1970), U.S.Code Cong. & Admin.News 1970, p. 5357, (Joint Appendix (JA) 322). Vehicle manufacturers must provide both a design- and-defect warranty established by section 207(a)5 and a performance warranty to be implemented under section 207(b). The short tests whose validity is challenged here are a part of the performance warranty scheme.
Under the performance warranty scheme, Congress directed the Administrator to promulgate two sets of regulations. First, he was to establish “short tests” for use in state and local inspection and maintenance programs so that vehicles which failed to conform to emission standards could be [243]*243identified. In other words, the short tests are intended to determine whether in-use vehicles would fail the Federal Test. Second, once facilities were available to conduct the short tests, the Administrator was to promulgate regulations requiring the manufacturer, instead of the vehicle owner, to pay for such repairs as would bring the vehicle into compliance with applicable emission standards.6
Because failure of a short test will trigger performance warranty liability on the part of a manufacturer, Congress established three prerequisites to the promulgation of short test regulations. EPA must determine that:
(1) testing methods and procedures are “available” to detect the failure of in-use vehicles to conform to emission standards;
(2) the tests are consistent with “good engineering practices”; and
(3) the tests are “reasonably capable of being correlated with” the test procedures utilized in certification testing (i.e., the Federal Test).
42 U.S.C. § 7541(b)(1) — (iii). The gravamen of petitioner’s challenge is the alleged failure of EPA to meet these prerequisites.
II. Procedural History
Although section 207 of the Act was enacted in 1970, EPA did not formally propose short test regulations until 1977, when, after conducting an extensive series of tests, the agency noticed a proposed rule.7 The notice proposed five short tests, the nature of which ranged from simple tests already in use in some inspection and maintenance programs to complex tests that had not yet actually been used. Proposed Rule, supra note 7, at 26,742-43 (JA 12). Because the data gathered in preparing the proposed rules were generated solely by testing vehicles under controlled laboratory conditions, the agency contemplated incorporating the effects of real world variables, such as ambient conditions, into numerous short test standards, or “cutpoints,” for particular classes of new vehicles on a yearly basis. Id. at 26,743, col. 1-2 (JA 2). “Cutpoints” are particular numerical values or “levels” for determining whether vehicles pass or fail the short test: a vehicle with emissions above the cutpoint fails the short test. EPA noted that short tests were subject to two types of incorrect predictions. An error of commission would occur if the test predicted a vehicle would fail the certification test when it would actually pass; an error of emission would occur if the test incorrectly predicted that a vehicle would pass the certification test when it would actually fail. Id. at 26,744, col. 1-2 (JA 3). Errors of commission can cause manufacturers to bear the costs of repairs for vehicles actually in compliance with emission standards, while errors of omission result in a loss of air quality benefits.
Nearly one hundred individuals and corporations, including all of the major American and import automobile manufacturers, submitted comments. On December 21, 1979, petitioner Motor Vehicle Manufacturers Association of the United States, Inc. (the “association”) petitioned the agency to repropose the short test regulations, chiefly on the ground that new studies, data, and technology cast doubt on the capability of the proposed tests to meet the statutory prerequisites. See Motor Vehicle Manufacturers Association’s Petition for Reproposal at 3 (JA 67). Six days later, EPA released a technical report8 containing new informa[244]*244tion, principally data from a study of motor vehicle emission testing procedures conducted for EPA in Portland, Oregon (hereinafter “the Portland Study”). A month after releasing the report, EPA noticed the reopening of the short-test comment period for thirty days9 to allow interested persons to comment on the data and analyses that had been generated since the close of the original comment period.10
Unsatisfied, the association filed a Supplement to its Petition for Reproposal,11 contending that the 30-day period for additional comments was too short to permit adequate analysis of the EPA Technical Report. The association also protested that regulations based on the Report necessarily would be so different and would raise so many new issues that reproposal was required. Nonetheless, petitioners and other interested parties submitted comments on the EPA Technical Report and on the Portland Study data.12
The Administrator promulgated the final rules at issue here on May 22, 1980. Short Test Regulations, supra note 3 (JA 224). EPA adopted test methods and procedures that were greatly simplified as compared to the proposed tests.13 Acknowledging that the short tests could not detect all excess emissions, the agency determined not to delay the entire program until more complex test procedures could be developed. Id at 34,810-11 (JA 233-34). Instead, and in order to ensure that the simplified tests adopted would “reasonably correlate” with the Federal Test Procedure, EPA developed cutpoints which will result in a percentage of incorrect test failures (errors of commission) no higher than would result if the Federal Test itself were used as the inspection test. While under this approach, it became — to the manufacturers’ advantage — somewhat easier to pass the short test than under the proposed approach, the agency concluded it was better to detect only the grossest emitters than to detect none at all. In this way, EPA sought to strike a balance between the interests of manufacturers and the public which would effectuate the intent behind passage of section 207(b).
This petition followed.
III. Standard op Review
Section 10(e) of the Administrative Procedure Act (APA), 5 U.S.C. § 706 (1976), governs the standard of review by this court.14 Asarco, Inc. v. EPA, 578 F.2d 319 [245]*245(1978). Under the APA, we must uphold the Administrator’s action in promulgating the short test regulation unless we find his action “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706. Agency action is arbitrary and capricious if the agency has failed to meet statutory, procedural, or constitutional requirements, Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 414, 91 S.Ct. 814, 822, 28 L.Ed.2d 136 (1971), or if its action is not supported by substantial evidence. Lead Industries Association, Inc. v. EPA, 647 F.2d 1130, 1145-48 (D.C.Cir.), cert. denied, 449 U.S. 1042, 101 S.Ct. 621, 66 L.Ed.2d 503 (1980); Pacific Legal Foundation v. Department of Transportation, 593 F.2d 1338, 1343 n. 35 (D.C.Cir.), cert. denied, 444 U.S. 830, 100 S.Ct. 57, 62 L.Ed.2d 38 (1979). The standard is “highly deferential” and presumes the validity of agency action. Citizens to Preserve Overton Park, Inc. v. Volpe, supra, 401 U.S. at 415-16, 91 S.Ct. at 823-24.
IV. The Statutory Prerequisites
The three statutory prerequisites to promulgation of the short test regulations are set out in section 207(b) of the Act:
If the Administrator determines that (i) there are available testing methods and procedures to ascertain whether, when in actual use throughout its useful life ..., each vehicle and engine to which regulations under section 202 apply complies with the emission standards, (ii) such methods and procedures are in accordance with good engineering practices, and (iii) such methods and procedures are reasonably capable of being correlated with tests conducted under section 206(a)(1) [i.e., the Federal Testing Procedure], then—
(1) he shall establish methods and procedures by regulation ....
42 U.S.C. § 7541(b)(i) — (iii). Petitioners contend that EPA’s determination that its short tests meet these three prerequisites is erroneous and thus arbitrary and capricious. We consider each prerequisite in turn.
A. Availability
The first prerequisite to promulgating short test regulations is a finding that “there are available testing methods and procedures to ascertain whether, in actual use throughout its useful life ..., each vehicle and engine to which regulations under section 202 apply complies with the emission standards.” Id. § 7541(b)(i) (emphasis added). The Administrator made such determination, concluding that the three tests adopted can reasonably be expected to serve their intended function of identifying vehicles that would fail to comply with emissions standards. Short Test Regulations, supra note 3, at 34,802, col. 3 (JA 225). Moreover, the agency found that all three tests were already being used in several state inspection and maintenance programs. Id. at 34,812, col. 2 (JA 235).
Petitioners assail the “availability” of the short tests on two grounds: short tests are not capable of determining compliance with emissions standards, and they do not measure NOx. Neither ground has merit.
1. Determining Compliance
Petitioners seize upon the agency’s concession that “[n]one of the short tests investigated by the Agency can reliably predict compliance with emissions standards.” Id. at 34,813, col. 2 (JA 236). See also id. at 34,803, col. 2 (JA 226) (“Such a requirement certainly cannot be met at this time, if it can be met at all.”). Because the statute, on its face, calls for short tests that will indicate “compliance] with ... emissions standards,” 42 U.S.C. § 7541(b)(i), petitioners contend that short tests must be able to predict whether a vehicle would pass and fail the Federal Test. Consequently, petitioners view the Administrator’s concession as dispositive. See Joint Brief for Petitioners and Intervenors at 23-24.
We cannot agree. While section 207(b) could be read as sanctioning only those methods that establish a vehicle’s ability, rather than inability, to pass the Federal Test, the Administrator has interpreted the statute as calling for tests which can identify failing vehicles with reasonable [246]*246accuracy so that manufacturers could be required to repair them. Short Test Regulations, supra note 3, at 34,812, cols. 1-2 (JA 235). It is well established that considerable deference is owed an agency’s interpretation of the statute it administers. E.g., Whirlpool Corp. v. Marshall, 445 U.S. 1, 11, 100 S.Ct. 883, 890, 63 L.Ed.2d 154 (1980); E.I. duPont de Nemours & Co. v. Train, 430 U.S. 112, 134-35, 97 S.Ct. 965, 978-79, 51 L.Ed.2d 204 (1977); Train v. NRDC, 421 U.S. 60, 75, 95 S.Ct. 1470, 1479, 43 L.Ed.2d 731 (1975). The Administrator properly rejected petitioners’ construction of section 207(b).
First, to read section 207(b)(i) as requiring that short tests be able to predict not only vehicles that would fail the Federal Test but also those that would pass it is to read the provision as establishing the requisite correlation that must be achieved between the Federal Test Procedure and the short tests. Petitioners’ argument, therefore, reduces to a reiteration of its argument, discussed infra Part IV C, that short tests must be classically correlated with the Federal Test Procedure. Section 207(b)(i), however, speaks of availability, not correlation; and the degree of correlation required of the short tests is established by section 207(b)(iii), not section 207(b)(i). Petitioners’ construction would thus render section 207(b)(iii) useless.
Second, petitioners’ construction would make it impossible to establish any short tests for the foreseeable future. See Short Test Regulations, supra note 3, at 34,803, col. 2 (JA 226). Such a result would frustrate the intent of Congress, which sought “to speed up, expand, and intensify the war against pollution” in enacting the Clean Air Act. H.R.Rep. No. 91-1146, 91st Cong., 2d Sess. 1 (1970) (emphasis added), U.S.Code Cong. & Admin.News 1970, p. 5356. See also id. at 5 (Progress-in controlling air pollution “has been regrettably slow.”). A statute should ordinarily be read to effectuate its purposes rather than to frustrate them. National Petroleum Refiners Association v. FTC, 482 F.2d 672, 689 (D.C.Cir.1973), cert. denied, 415 U.S. 951, 94 S.Ct. 1475, 39 L.Ed.2d 567 (1974). The Administrator’s construction does so, and we find it reasonable.
2. The Failure to Measure NOx
The Federal Test Procedure measures three pollutants for which EPA has established emissions standards: HC, CO and NOx. See Emission Standards for 1981 Light-Duty Vehicles, 40 C.F.R. § 86.081-8 (1982). EPA concedes that its short tests cannot detect excessive NOx emissions. Short Test Regulations, supra note 3, at 34,811, col. 1 (JA 234). Petitioners argue that this concession indicates short tests are not “available” to ascertain compliance with emissions standards.
Nothing in section 207(b)(i), or in the'rest of the Act, suggests that short tests cannot be implemented until excessive emissions of NOx can also be detected. Nor can petitioners advert to anything in the policy rationale or legislative history of the Act which might indicate that the agency must abjure two-thirds of a loaf just because the whole loaf is not yet available. While we are moved by petitioners’ concern that emissions of all pollutants be policed effectively, we must reject an interpretation which frustrates the intent of Congress by delaying indefinitely the establishment of any short tests whatsoever. See United States v. Chrysler Corp., 591 F.2d 958, 961 (D.C.Cir.1979); National Petroleum Refiners Association v. FTC, supra, 482 F.2d at 689.
The short tests are “available” within the meaning of section 207(b)(i).
B. Good Engineering Practices
Section 207(b)(ii) requires that short test methods and procedures be “in accordance with good engineering practices.” 42 U.S.C. § 7541(b)(ii). The Administrator determined that a short test accords with good engineering practices if it is capable of detecting vehicles with grossly excessive emissions. See Short Test Regulations, supra note 3, at 34,802, col. 3, 34,812 — 13 (JA 225, 236). Because testing of almost 6,000 vehicles from different locations (in some [247]*247cases from actual inspection and maintenance lanes) showed that high-emitting vehicles were accurately identified, the Administrator found that the adopted short tests have demonstrated this capability. See id. at 34,802, col. 3 (JA 225).
Undaunted, petitioners launch three distinct attacks on the agency’s finding that the short tests satisfy the “good engineering practices” prerequisite: (1) EPA collected no data on light-duty trucks or light-duty diesel vehicles; (2) EPA’s data are obsolete; and (3) EPA failed to investigate the effect of ambient conditions on correlation. All three attacks fail to provide a basis for overturning the short test regulations.
1. Light-Duty Trucks and Diesels
Petitioners assert that the short tests and cutpoints cannot be applied to 1981 diesel-powered light-duty vehicles (“diesels”) or to 1981 gasoline-powered or diesel-powered light-duty trucks (“light trucks”) because their emission control systems are different from those of the gasoline-powered automobiles tested in the Portland Study and because they are subject to different emissions regulations. Joint Brief of Petitioners and Intervenor at 12-14.
EPA concluded that the Portland Study automobile data were applicable to light trucks and diesels because the vehicles tested were “adequately representative of all vehicles to be covered by the rule.” Short Test Regulations, supra note 3, at 34,824, col. 3 (JA 247). Petitioners rejoin that automobile emissions systems “are not identical and do not operate on identical vehicles” — an observation we should have thought implicit in the words “adequately representative.” Joint Brief of Petitioners and Intervenors at 13 (emphasis added). The emphasis on differences between diesel and gasoline vehicles ignores the fact that both types will be subject to identical emissions standards for HC and CO in 1981 and later model years. Light-duty diesel manufacturers themselves have told the Administrator that diesels are inherently cleaner than gasoline vehicles for HC and CO.15 Moreover, diesels emit very low levels of HC and CO under short test conditions because they idle with very lean fuel/air mixtures. Short Test Regulations, supra note 3, at 34,806, col. 1 (JA 229). As a result, cutpoints developed to identify grossly excessive emitters among gasoline vehicles will identify relatively fewer diesels. Since it is to the manufacturers’ advantage that fewer diesels will fail a short test, it ill behooves petitioners to complain that EPA has failed to quantify whatever difference inheres in the emissions of gasoline and diesel vehicles.
As to light trucks, the Administrator reasonably concluded that the testing of 1981 trucks would achieve the same level of correlation as was achieved for the Portland passenger cars. Light trucks in 1981 and later model years will be using the same emissions control technologies as those used by the 1977 cars tested in the Portland Study. See id. at 34,824, col. 3 (JA 247). The two types of vehicles are treated differently only in the sense that 1981 trucks are subject to less stringent HC and CO standards than 1981 passenger cars. See id. at 34,805, col. 1 (JA 228). This indicates how conservative the cutpoints are when applied to 1981 and later model automobiles, whose technologies will permit emission of much lower HC and CO levels than did the Portland vehicles. See id. at 34,811, cols. 2-3 (JA 234). Again, any slight discrepancies favor manufacturers.
2. The Alleged Obsolescence of EPA Data
Petitioners complain that changes in emissions control technology made by manufacturers since the proposed rule was noticed have rendered obsolete the data on which EPA based its finding that the short tests adopted accord with good engineering [248]*248practices. Such “obsolescence” is, of course, virtually inevitable under a statutory scheme that requires detailed technical findings and elaborate procedural protections. The short answer to petitioners’ argument is that any technological improvements are to their advantage: as emissions control systems improve, the number of short test failures will of necessity decrease. Petitioners’ construction of section 207(b)(ii) would never permit the agency to establish short tests.
3. The Effect of Ambient Conditions
Petitioners allege that variations in test temperature, humidity, and atmospheric pressure16 will make correlation between a short test and the Federal Test Procedure impossible. EPA determined, however, that “[t]he effects of ambient temperature and fuel are less pronounced and are not deemed to be significant at the level of standards adopted.” Id. at 34,817, col. 1 (JA 240) (emphasis added). • In other words, as petitioners phrase it, the agency has factored in ambient variables by the “manipulation of short test cut-points.” Joint Brief of Petitioners and Intervenor at 19. Petitioners do not deny that such “manipulation” reduces to insignificance any effects ambient conditions might have on manufacturers’ liability under the performance warranty. Rather, they dispute whether use of cutpoints to fudge complex technological problems can itself be considered “good engineering practice.”
We believe such “manipulation” is permissible, so long as it is applied with internal consistency, effectuates the intent of Congress, and does not saddle manufacturers with'significantly greater liability under the performance warranty scheme than would result if more perfect correlation with the Federal Test Procedure could be achieved. The Administrator’s resolution of thorny technological problems through the use of cutpoints represents a reasonable accommodation of the conflicting interests entrusted to his care. In this respect, we are mindful of the Supreme Court’s recent admonition that when an agency makes determinations “within its area of special expertise ... at the frontiers of science,” “a reviewing court must be at its most deferential.” Baltimore Gas & Electric Co. v. NRDC, - U.S. -, -, 103 S.Ct. 2246, 2256, 76 L.Ed.2d 437 (1983).
C. Reasonable Correlation with the Federal Testing Procedure
Section 207(b)(iii) requires that EPA determine that its short tests “are reasonably capable of being correlated with tests conducted under section 206(a)(1)” — i.e., with the Federal Test Procedure. 42 U.S.C. § 7541(b)(iii) (emphasis added). Petitioners argue that the Administrator erred in adopting short tests which do not correlate with the Federal Test Procedure in the “ ‘classical statistical sense.’ ”17 Joint Brief of Petitioners and Intervenor at 19 (quoting Short Test Regulations, supra note 3, at 34,813, col. 2 (JA 236)). This contention is unsupportable, as the statute, by its own terms, requires “reasonable” not classical correlation. Had Congress intended to prescribe classical correlation it could have done so with less ambiguity. Furthermore, petitioners can refer to nothing in the legislative history which might suggest that the plain language of the statute does not express the meaning Congress intended to convey.
Alternatively, petitioners contend that the correlation demonstrated by EPA is not “reasonable” in any event. EPA first calculated the variation in results obtained from performing successive Federal Testing Procedures on the same vehicle. Upon determining that 5.3 percent of vehicles which pass the first Federal Test will fail a second one, .EPA concluded that a short test which [249]*249has an error-of-commission rate when compared with the Federal Test of less than 5.3 percent “reasonably” correlates with the Federal Testing Procedure. See EPA Technical Report, supra note 8, at 8-11 (JA 47-50). By changing short-test cutpoints to ensure that the error-of-commission rate will not exceed 5.3 percent, the agency can identify vehicles which emit high levels of HC and CO without exposing manufacturers to greater liability under the performance warranty than would result if the short tests achieved perfect correlation with the Federal Testing Procedure. In fact, the agency’s procedure ensures that manufacturers’ exposure will be less than under a procedure which achieved perfect correlation:
By varying the severity of the short test cut point (i.e., the numerical value used to predict passing or failing the Federal Test) it is possible to reduce errors of commission to any desired level, but always at the cost of increasing errors of omission.
Proposed Rule, supra note 7, at 26,744, col. 2 (JA 3) (emphasis added).
Petitioners denigrate this technique as a “teeter-totter approach” which is “intrinsically defective” and thus arbitrary and capricious. Joint Brief of Petitioners and Intervenor at 21. We find, however, that the method assures “reasonable” correlation with the Federal Testing Procedure within the meaning and aims of section 207(b). Congress recognized that the primary difference between the “pre-certification test procedure” (i.e., the Federal Testing Procedure) and the “quick test” was that the latter would have to be a test which would be applied during the normal time for á regular vehicle inspection, while providing a method to detect individual vehicle noncompliance. See S.Rep. No. 91-1196, 91st Cong., 2d Sess. 28 (1970). Obviously, Congress did not intend to require the impossible task of developing a “quick” test with the accuracy of the Federal Testing Procedure; rather, it wanted to ensure that an accommodation was reached among competing goals of consumer convenience, improved air quality, and the technical accuracy which would ensure that manufacturers are not forced to “repair” significant numbers of properly functioning vehicles.
The Administrator has fashioned a reasonable compromise. In fact, as noted, any imperfections in the correlation achieved between the short tests and the Federal Testing Procedure inure to the benefit of the manufacturers. To that extent, in chiding the Administrator for failing to eliminate those imperfections, petitioners are reaching beyond their own concerns. We decline to adopt a vapid construction of section 207(b) which would frustrate — perhaps forever — effectuation of the statutory design. See National Petroleum Refiners Association, supra, 482 F.2d at 689.18
V. Conclusion
We hold that the Administrator has met the statutory prerequisites of section 207(b) for promulgating final short test regulations.19 Accordingly, the petition is dismissed.
Judgment accordingly.