WILLIAMS, Senior Circuit Judge:

Title II of the Clean Air Act, 42 U.S.C. § 7521 et seq. (1955), sets up a program for the regulation of both motor vehicles and their fuels in order to reduce harmful emissions. Section 206 charges the Environmental Protection Agency with testing new motor vehicles to ensure that each vehicle's emissions will comply with federal emissions standards throughout its "useful life." 42 U.S.C. § 7525(a)(1). Section 206(d) says that the agency "shall by regulation establish methods and procedures for making tests under this section." Id. § 7525(d) (emphasis added).

In a rulemaking pursuant to § 206, the EPA adopted a Compliance Assurance Program or "CAP 2000." 64 Fed. Reg. 23,906 (1999). CAP 2000 does not, however, set out "methods and procedures for making tests." Rather, it establishes a framework for automobile manufacturers to develop their own tests, to be used once the EPA gives approval, case-by-case, after private proceedings with each manufacturer.

Petitioner Ethyl manufactures and markets fuel and lubricant additives for use in motor vehicles. It argues that CAP 2000 violates the Act because it provides for test procedures and methods to be vetted in individual closed proceedings rather than in a notice-and-comment rulemaking. And it claims to be injured because the mechanism adopted by the EPA deprives it of the opportunity to observe the rulemaking process and thus gain information useful in its efforts both to develop and improve its products and to key them to the certification tests. For the reasons given below we grant the petitions.

* * *

Before a manufacturer may introduce a new motor vehicle into commerce, it must obtain an EPA certificate indicating compliance with the requirements of the Act and applicable regulations. It submits an application containing test data and other information specified by the EPA, which issues a certificate if the manufacturer has shown, among other things, that the vehicle's emissions control systems will achieve compliance with emissions standards over the vehicle's full useful life. See 40 CFR § 86.1848-01.

Critical here is the question of the control systems' possible deterioration over time. Before 1993 EPA had had a durability test that called for prototype vehicles to be driven over a 50,000-mile course known as the Automobile Manufacturers Association ("AMA") driving cycle. 58 Fed. Reg. 3994, 3995/1 (1993). In 1993 it adopted a "revised durability program" or "RDP" that retained that test "as the standard EPA-defined procedure." Id. But the RDP regulations also permitted automobile companies to develop alternative test methods and procedures provided that they (a) obtained EPA approval for each such test and (b) performed in-use testing to verify the accuracy of the emissions deterioration predictions made by their tests. See id. at 3995. The EPA did not adopt these tests through rulemaking but simply approved them on a case-by-case basis.

In May 1999 the EPA replaced RDP with CAP 2000. These regulations eliminate the AMA driving cycle as an EPA-defined test method. Instead, the program available as an alternative in 1993-99, under which manufacturers are to develop their own emissions durability test methods and procedures, has become the sole method. Thus, rather than promulgating methods and procedures for durability testing itself, the EPA now requires, through CAP 2000, that "[t]he manufacturer shall propose" a durability program. 40 C.F.R. § 86.1823-01. Each manufacturer is required to obtain EPA approval for its tests, and must verify its results through in-use testing.

Manufacturer-proposed tests under CAP 2000 must (a) "effectively predict the expected deterioration of candidate in-use vehicles over their full and intermediate useful life," and (b) be "consistent with good engineering judgment." 40 C.F.R. § 86.1823-01(a). Within these criteria, the manufacturer-developed mileage accumulation procedures are to be

based upon whole-vehicle full-mileage accumulation, whole-vehicle accelerated mileage accumulation (e.g., where 40,000 miles on a severe accumulation cycle is equivalent to 100,000 miles of normal in-use driving), bench aging of individual components or systems, or other approaches approved by the Administrator.

40 C.F.R. § 86.1823-01(a)(1)(ii). The "bench aging" referred to is a system whereby components are removed from the vehicle and tested for durability separately. 40 C.F.R. § 86.1823-01(a)(1)(B).

In adopting this system of individualized test approval, the EPA explicitly found that "rulemaking for each durability program is not required." 64 Fed.Reg. at 23,914/3. It also said, in a response to petitions for reconsideration by Ethyl, that public participation in the certification process would interfere with the process of reviewing manufacturers' submissions "because of the large amount of information claimed confidential" and that, because the process was annual, the use of notice-and-comment procedures would be "administratively burdensome." August 23, 2001 Response to Ethyl Corporation Petitions Denying Reconsideration of Three EPA regulations: CAP 2000, Heavy Duty Gasoline, and OBD/IM, EPA Air Docket A-96-50, No. VI-C-03, 39.

Ethyl challenges not only CAP 2000 but also regulations governing the certification of heavy duty vehicles and engines, 65 Fed. Reg. 59,896 (2000), which incorporate the CAP 2000 regulations by reference, and EPA's denial of its various petitions for reconsideration, 66 Fed. Reg. 45,777 (2001).

* * *

The EPA argues that we need not reach the merits because, it says, Ethyl lacks both Article III and "prudential" standing. We think it has both.

As is well known, Article III requires a party seeking judicial relief to show (1) that it has suffered an "injury in fact"; (2) that the injury is caused by or fairly traceable to the challenged actions of the defendant; and (3) that it is likely that the injury will be redressed by a favorable decision. Lujan v. Defenders of Wildlife, 504 U.S. 555, 560-561, 112 S.Ct. 2130, 2136, 119 L.Ed.2d 351 (1992).

Ethyl's assertions of injury fall into two categories. First, as we've already mentioned, it says that as a manufacturer of additives for motor vehicle fuels it has an interest in understanding the test methods and procedures by which the EPA certifies new motor vehicles. CAP 2000's provision for closed-door adoption of emission test procedures deprives Ethyl of information that might well help it develop and improve its products with an eye to conformity to emissions needs.

Second, Ethyl says that CAP 2000 deprives it of information that might be useful for securing EPA approval for its own fuel additive products under the Act. It points in particular to § 211(f), which prohibits use of any fuel or fuel additive that is not "substantially similar" to the fuels used to certify vehicles under § 206, unless a waiver is obtained from EPA. See 42 U.S.C. § 7545(f).

EPA's response to this focuses almost entirely on Ethyl's asserted interest in the § 211(f) waiver process, completely ignoring its interest in obtaining information about vehicle certification for present-day research and development of products that will be judged (by both the government and consumers) according to their effect on vehicle emissions. The Supreme Court has made clear, however, that a denial of access to information can work an "injury in fact" for standing purposes, at least where a statute (on the claimants' reading) requires that the information "be publicly disclosed" and there "is no reason to doubt their claim that the information would help them." Federal Election Comm'n v. Akins, 524 U.S. 11, 21, 118 S.Ct. 1777, 1784, 141 L.Ed.2d 10 (1998). Here, against Ethyl's fairly detailed description of how the information that open rulemaking proceedings provide would prove useful to it, the EPA offers little more than a vague shrug of skepticism.

Because Article III standing is clear from Ethyl's informational and market interests in the vehicle-testing program, we need not address the interest based on its need to seek variances under § 211(f).

To show "prudential" standing, Ethyl must fall within the "zone of interests" protected or regulated by the Act. See Bennett v. Spear, 520 U.S. 154, 162, 117 S.Ct. 1154, 1161, 137 L.Ed.2d 281 (1997). The test is not a particularly demanding one, Clarke v. Securities Industry Association, 479 U.S. 388, 399, 107 S.Ct. 750, 757, 93 L.Ed.2d 757 (1987), and includes not only those challengers expressly mentioned by Congress, but also unmentioned potential challengers that Congress would have thought useful for the statute's purpose (whose challenges thereby support an inference that Congress would have intended eligibility). See Hazardous Waste Treatment Council v. EPA, 861 F.2d 277, 283 (D.C. Cir. 1988). It excludes parties whose interests are not consistent with the purposes of the statute in question. See Clarke, 479 U.S. at 399, 107 S.Ct. at 757. As a manufacturer of fuel additives seeking an open process for testing the emissions control systems whose character may affect the efficacy of its products, Ethyl's interests appear congruent with those of the statute, i.e., the development of products that will reduce harmful air pollutants. Indeed, this court has long recognized the interdependence between motor vehicle certification under the Act (the process at stake here) and fuel regulations (under which Ethyl is a direct regulatee). See e.g., Lubrizol Corp. v. EPA, 562 F.2d 807, 810 (D.C. Cir. 1977); Amoco Oil Co. v. EPA, 501 F.2d 722, 737 (D.C. Cir. 1974). The case is not unlike National Cottonseed Products Association v. Brock, 825 F.2d 482, 489-492 (D.C. Cir. 1987), where we found standing for a manufacturer whose respirators had been assigned a low rating by an agency supervising conditions in a workplace for which the respirators were a potential means of compliance. We treated the respirator seller's interest, and that of the regulated firms, as "`two sides of the same coin.'" Id. at 491 (quoting FAIC Sec., Inc. v. United States, 768 F.2d 352, 359 (D.C. Cir. 1985)).

On to the merits: As we said, § 206(d) of the Act states that the administrator "shall by regulation establish methods and procedures for making tests under this section." 42 U.S.C. § 7525(d). Although special provisions govern review under the Act, here the relevant provisions are the same as under the Administrative Procedure Act. We are to reverse the challenged EPA actions if they are "arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law" or "in excess of statutory jurisdiction, authority, or limitations." 42 U.S.C. § 7607(d)(9)(A), (C).

CAP 2000 does not, as § 206 directs, "establish methods and procedures for making tests," and it is the only "regulation" in the picture. Instead, it provides criteria for individual automobile manufacturers to develop their own test methods and procedures, which the EPA approves in a process that does not involve rulemaking.

Conceivably § 206(d)'s requirement that EPA use regulation to "establish methods and procedures for making tests" could be squared with the record by reading "making tests" as referring to devising the tests rather than conducting them. Thus Congress would be mandating that the EPA use regulations merely to set up a system for picking tests (which might then be picked any old way) rather than mandating the use of regulations to decide how the tests themselves should be conducted. But nothing in the context of the provision suggests that the "establish[ment]" under § 206(d) is to be so remote from the actual process of conducting tests, and it is hard to see any congressional purposes that would be served by such a requirement. Indeed, neither in the administrative proceedings nor before us has the EPA invoked such a reading.

Rather, the EPA seeks to defend CAP 2000 by treating the issue as involving simply the level of specificity or generality at which it was supposed to act, citing American Trucking Associations v. Department of Transportation, 166 F.3d 374 (D.C. Cir. 1999) (agency to promulgate by regulation safety rating "requirements" and means to determine whether carriers had met the requirements), and New Mexico v. EPA, 114 F.3d 290 (D.C. Cir. 1997) (agency to promulgate "criteria" for a certification process). In those cases, as Congress had not specified the level of specificity expected of the agency, we held that the agency was entitled to broad deference in picking the suitable level. See American Trucking, 166 F.3d at 379-80; New Mexico, 114 F.3d at 294. But here Ethyl's challenge is not that the EPA was too general in establishing test procedures by regulation, but that it didn't establish them by regulation at all.

EPA's failure to act by regulation is thus similar to, and controlled by, our decision in MST Express v. Department of Transportation, 108 F.3d 401 (1997), which preceded American Trucking and involved the same statutory requirement of proceeding by regulation in setting safety requirements for common carriers. Rather than promulgate regulations stating the means for determining whether carriers met the safety fitness requirements, the agency had simply required a carrier to "demonstrate that it has adequate controls in place" to ensure compliance with the substantive requirements, and had developed a "safety fitness rating methodology." Id. at 402, 403. This methodology provided agency inspectors with detailed guidelines for evaluating a motor carrier's safety rating — but it was not the product of notice-and-comment rulemaking. See id. at 403. We found that the agency had "failed to carry out its statutory obligation to establish by regulation a means of determining whether a carrier has complied with the safety fitness requirements." Id. at 406. EPA's error here is similar.

There may, of course, be cases in which it is hard to distinguish between promulgations of (1) vaguely articulated test procedures (which would be reviewed deferentially under such cases as American Trucking) and (2) procedures for later development of tests (invalid under MST Express). Both, after all, necessarily imply a later (or at least different) proceeding in which the agency will fill in details. In this case, however, one can distinguish on the basis of the language used by the agency. With CAP 2000, the EPA does not claim to have itself articulated even a vague durability test. Rather, CAP 2000 requires that "[t]he manufacturer shall propose a durability program" for EPA approval. 40 C.F.R. § 86.1823-01(a). It thus falls on the forbidden side of the line.

The EPA also defends CAP 2000 on grounds that seem to flout the evident congressional purpose. First, it argues that because it has chosen to approve test procedures only for one model year at a time, proceeding by regulation would be administratively burdensome. Obviously this cannot overcome a clear congressional command. Further, it is true only in the sense that an open procedure — the very thing mandated by Congress — is less convenient than a closed one. It may be. Other parties may raise questions or find fault in procedures that look fine to the agency and the auto makers. But Congress has already made the trade-off. Nothing in our opinion requires that EPA use only a "one-size-fits-all" test method. All that is required is that it establish its procedures, no matter how variegated, "by regulation."

Finally, both EPA and the auto manufacturers who intervene on its behalf argue that the approach of CAP 2000 is necessary because of the presence of what the manufacturers believe to be "confidential business information" ("CBI"). If the EPA were to establish test methods and procedures by regulation, they say, important CBI might become public, allowing competitors to "back engineer" their products. See Tr. of Oral Argument at 29-30. Moreover, they argue that the sheer "amount of claimed confidential business information would significantly reduce the usefulness of public notice and an opportunity to comment upon manufacturers' durability programs." EPA Br. at 44.

It is hard to know what to make of this argument. First and foremost, § 208(c) provides that the administrator may protect the confidentiality of "methods or processes entitled to protection as trade secrets." 42 U.S.C. § 7542(c). Especially given this available remedy, it seems to us a complete non sequitur to suggest that because a procedure (the rulemaking mandated by § 206(d)) may involve some protectable CBI, the entire procedure should be short-circuited and replaced with a cluster of closed bargaining sessions between the EPA and each manufacturer. Congress obviously expected that rulemakings would proceed despite the existence of CBI that would require protection under § 208(c). Plainly the theory provides no basis for disregarding the congressional command.

* * *

CAP 2000, rather than constituting an EPA establishment "by regulation" of "methods and procedures for making tests," as required by § 206(d), is instead a promulgation of criteria for the later establishment of such methods and procedures by private negotiation between the EPA and each regulated auto maker. So it is "not in accordance with law." We therefore vacate the CAP 2000 program and remand the case to the EPA with instructions to establish test methods and procedures by regulation.

So ordered.