Appellants Peter Maier, the Intermountain Water Alliance, the Atlantic States Legal Foundation, the Utah Wilderness Association, and Kay Henry1 petitioned the Environmental Protection Agency (EPA) to initiate rulemaking under the Clean Water Act (CWA), 33 U.S.C. §§ 1251-1387. Mr. Maier contended that recent developments in municipal wastewater technology have rendered the EPA’s regulations for secondary treatment inadequate, and therefore the EPA must promulgate new standards. The EPA denied the petition, and Mr. Maier appealed to this court. We affirm.
I.
A.
We start with an overview of the relevant statutory scheme. The CWA aims “to restore and maintain the chemical, physical, and biological integrity of the Nation’s waters” by reducing and eventually eliminating the discharge of pollutants. 33 U.S.C. § 1251(a), (a)(1). “[T]he basic structure of the [CWA] ... translates Congress’ broad goal of eliminating ‘the discharge of pollutants into the navigable waters’ into specific requirements that must be met by individual point sources.” EPA v. National Crushed Stone Ass’n, 449 U.S. 64, 69, 101 S.Ct. 295, 300, 66 L.Ed.2d 268 (1980) (quoting 33 U.S.C. § 1251(a)(1)) (citations omitted). A “point source” is defined as “any discernible, confined and discrete conveyance ... from which pollutants are or may be discharged.” 33 U.S.C. § 1362(14). The CWA prohibits the discharge of any pollutant from a point source unless that discharge complies with the CWA’s requirements. 33 U.S.C. § 1311(a). Compliance can be achieved by obtaining a National Pollutant Discharge Elimination System (NPDES) permit, pursuant to 33 U.S.C. § 1342, which establishes technology-based controls and ensures compliance with state or federal water quality standards. 33 U.S.C. § 1311(b)(1)(C). These permits generally contain quantitative limits on the amounts of specified pollutants that may be discharged. See generally Oklahoma v. EPA, 908 F.2d 595, 597-98 (10th Cir.1990), rev’d on other grounds sub nom., Arkansas v. Oklahoma, 503 U.S. 91, 112 S.Ct. 1046, 117 L.Ed.2d 239 (1992).
The CWA mandates varying standards of technology-based treatment as the minimum requirement for different categories of point sources. 33 U.S.C. §§ 1311, 1314. Under section 1311, “effluent limitations for point sources, other than publicly owned treatment works [POTWs] ... require the application of the best practicable control technology currently available as defined by the Administrator.” Id. § 1311(b)(1)(A). By contrast, the same section requires “for publicly owned *1035treatment works ... effluent limitations based upon secondary treatment.”2 Id. § 1811(b)(1)(B). The EPA may supplement the minimum requirements by using individual permits to impose “any more stringent limitation ... necessary to meet water quality standards.” Id. § 1311(b)(1)(C). When Congress first enacted these provisions, it intended to phase in a requirement that publicly owned treatment works utilize “best practicable waste treatment technology,” a stricter standard than secondary treatment. 33 U.S.C. § 1311(b)(2)(B) (1973); see also S.Rep. No. 92-414, at 43 (1972), reprinted in 1972 U.S.C.C.A.N. 3668, 3709 (“Publicly-owned treatment systems must meet the secondary treatment requirement of Phase I and, in Phase II, the mandate requires the best practicable treatment____”). In 1981, however, Congress repealed this general requirement and limited the stricter standard to federally-funded POTWs. Municipal Wastewater Treatment Construction Grant Amendments of 1981, Pub.L. No. 97-117, § 21(b), 95 Stat. 1623,1632 (1981).
B.
Secondary treatment, which is the basic requirement for all POTWs, is principally concerned with removing biological pollutants which affect the oxygen content of wastewater. Healthy waters contain dissolved oxygen upon which flora and fauna rely, but biological pollutants “demand” and consume this oxygen. The rate at which dissolved oxygen is consumed is measured by a parameter called “biochemical oxygen demand” (BOD). BOD actually measures the effect of two components of oxygen depletion which, with proper testing, can be disaggregated: CBOD and NOD.3 The EPA’s regulations for standard secondary treatment have always focused on BOD, and in particular on control of CBOD.4
Initially, the EPA required testing for BOD by a standard five-day test of overall BOD levels, the so-called BOD5 test. See rec. at 121-22. But the EPA became concerned that the BOD5 test produced erroneous indications of effluent quality. Secondary treatment with existing technology5 under some operating conditions could lead to increased nitrification (NOD), and inflated BODb values, despite producing effluent of better quality than facilities with lower BOD5. Id.; Secondary Treatment Information, 48 Fed.Reg. 52,272, 52,275-6. In a 1984 rulemaking, the EPA addressed the problem by revising its regulations to allow *1036the permitting authority to require facilities to employ a more specific measure of CBOD rather than the general test for BOD. See Secondary Treatment Regulation, 49 Fed.Reg. 36,986, 36,988-90, 36,998-99 (1984) (to be codified at 40 C.F.R. pt. 133). In effect, EPA recognized that NOD can constitute a significant component of BOD levels, but concluded that measuring NOD in some cases might distort rather than enhance accurate assessment of effluent quality-
During this rulemaking, commenting parties, including Mr. Maier, suggested that secondary treatment standards ought to directly address NOD or ultimate BOD.6 See id. at 36,999; rec. at 297. The EPA concluded, however, that NOD levels were quite variable and were therefore inappropriate as a criterion for generally-applicable standards. The EPA maintained that NOD would be better dealt with on a ease-by-ease basis in NPDES permitting. Secondary Treatment Regulation, 49 Fed.Reg. at 36,999; rec. at 297-98. The EPA therefore characterized NOD controls as a form of “advance treatment” to be imposed by permit where necessary. The EPA also noted that total impact on dissolved oxygen level (ultimate BOD) is to be considered in the NPDES permitting process. Secondary Treatment Regulation, 49 Fed.Reg. at 36,999. The EPA’s 1984 rule-making was not appealed to the federal courts. Thus, current regulations for secondary treatment do not establish standards for NOD, although an individual facility’s NPDES permit may well impose requirements for NOD.
In 1993, Mr. Maier filed a petition requesting the EPA to initiate rulemaking to set parameters for NOD and ultimate BOD as part of its secondary treatment regulations. Mr. Maier argued that the existing regulations, setting parameters only for CBOD and BOD5, were inadequate because new secondary treatment technology made it feasible and cost-effective to control both CBOD and NOD. The EPA did not controvert that controlling NOD was now feasible and cost-effective for some (but not all) POTWs, but emphasized that the impact of NOD is highly variable and dependent upon such factors as temperature and rate of flow of the receiving water body. In the EPA’s view, this variability continues to justify its regulatory decision to control CBOD with a generally-applieable regulation, but to control NOD on a ease-by-case basis through the permitting process. Consequently, the EPA denied Mr. Maier’s petition. Mr. Maier filed the instant petition for review in this Court pursuant to 33 U.S.C. § 1369(b)(1).
II.
We first determine whether we have jurisdiction. Although both parties agree that we have jurisdiction, “no action of the parties can confer subject-matter jurisdiction upon a federal court.” Insurance Corp. of Ireland, Ltd. v. Compagnie des Bauxites de Guinee, 456 U.S. 694, 702, 102 S.Ct. 2099, 2104, 72 L.Ed.2d 492 (1982). We have an independent duty to examine our own jurisdiction. Lopez v. Behles, 14 F.3d 1497, 1499 (10th Cir.), cert. denied, 513 U.S. 818, 115 S.Ct. 77, 130 L.Ed.2d 31 (1994).
The CWA establishes a bifurcated system of judicial review. Section 1365 “confers jurisdiction on the federal district courts, not courts of appeal, to review any action ‘where there is alleged a failure of the Administrator to perform any act or duty under this Act which is not discretionary with the Administrator.’’’ Trustees for Alaska v. EPA, 749 F.2d 549, 558 (9th Cir.1984) (quoting 33 U.S.C. § 1365(a)(2)). Section 1369 of the CWA provides that “[rjeview of the Administrator’s action ... (E) in approving or *1037promulgating any effluent limitation or other limitation under section 1311,1812, or 1316 of this title ... may be had by any interested person in the Circuit Court of Appeals of the United States.” 33 U.S.C. § 1369(b)(1).7 The jurisdictional grant in section 1369 is exclusive. American Petroleum Inst. v. Train, 526 F.2d 1343, 1344 (10th Cir.1975). We must determine whether the EPA’s denial of Maier’s petition to institute rulemaking constitutes an “action ... in approving or promulgating any effluent limitation or other limitation,” and thus falls within the ambit of section 1369(b)(1).
The District of Columbia Circuit dealt with a similar problem in Oljato Chapter of Navajo Tribe v. Train, 515 F.2d 654 (D.C.Cir.1975), which interpreted jurisdictional provisions of the Clean Air Act (CAA) in the context of the EPA’s refusal to revise previously promulgated standards for certain power plants. As in the CWA, the CAA contained one section permitting citizen suits to be brought in federal district court, id. at 657-58 n. 4 (citing CAA § 304, 42 U.S.C.A. § 1857h-2(a) & (b) (1970)), and another granting jurisdiction to the courts of appeal to “review of action of the Administrator in promulgating” standards of performance, id. at 657 n. 3 (quoting CAA § 307(b), 42 U.S.C.A. § 1857h-5(b) (1975 pocket part)). The petitioners in Oljato Chapter argued that an EPA standard had been rendered obsolete by changes in technology, and that the Administrator’s failure to revise the rule “constituted a failure to perform a nondiscretionary duty, thereby conferring District Court jurisdiction.” Id. at 658. The court rejected petitioners’ attempts to distinguish “between judicial review of an original rule and review of a subsequent refusal to modify or reverse that rule,” noting that
[wjhile we have no difficulty in making the suggested distinction, we conclude that Congress intended all review related to the continuing validity of standards of performance to be included within the exclusive scope of [the section providing for appellate review]; that this action is premised upon the Administrator’s refusal to revise the standard is of no jurisdictional import.
Id. at 659-60. Oljato Chapter relied both on the CAA’s legislative history, and on the fact that
[s]ince a revision by the Administrator is the ultimate goal of a new information appeal, it makes little sense to suggest that this court is stripped of its ... jurisdiction whenever a party attempts to avoid litigation by first asking the Administrator voluntarily to make the same revision a successful ... petition would require.
Id. at 660.
As was the case in Oljato Chapter, the legislative history of the CWA speaks directly to the case at hand:
The Committee recognizes that it would not be in the public interest to measure for all time the adequacy of a promulgation of any standard requirement or regulation by the information available at the time of such promulgation. In the area of protection of public health and environmental quality, it is clear that new information will be developed and that such information may dictate a revision or modification of any promulgated standard, requirement, or regulation established under the act. The judicial review section, therefore, provides that any person may challenge any requirement after the date of promulgation whenever it is alleged that significant new information has become available.
S.Rep. No. 92-414, at 85 (1972), reprinted in 1972 U.S.C.C.A.N. 3668, 3751. Congress plainly intended that the Administrator’s refusal to institute rulemaking in the face of new information could be reviewed directly in a circuit court.
Section 1369’s language about an Administrator’s “action ... in approving or promulgating any ... limitation” arguably does not apply to the Administrator’s refusal to promulgate a rule in the first instance. We *1038agree with the court in Oljato Chapter, however, that a challenge to the refusal to revise a rule in the face of new information is more akin to a challenge to the existing rule than a challenge to the refusal to promulgate a new rule.8 Because Mr. Maier is essentially challenging the sufficiency of the EPA’s secondary treatment regulation, we have no difficulty construing this as a challenge to an “action in approving or promulgating” under section 1369. Where petitioners’ challenge is to the substance of a regulation that the agency has already promulgated, exclusive jurisdiction in the court of appeals may not be evaded merely by styling the claim as one for failure to revise. Cf. David P. Currie, Air Pollution-. Federal Law and Analysis § 9.10, at 9-31 (1981) (“In short, allegations that the Administrator has failed to take action required by statute should not be permitted to circumvent the plain statutory command that judicial review of decisions respecting implementation plans and other regulations is to be in the courts of appeals____”). “Absent a far clearer expression of congressional intent, we are unwilling to read the CWA as creating such a seemingly irrational bifurcated system.” Crown Simpson Pulp Co. v. Costle, 445 U.S. 193, 196-97, 100 S.Ct. 1093, 1095, 63 L.Ed.2d 312 (1980) (rejecting attempt to distinguish challenge to EPA veto of a state permit from a challenge to EPA issuance of a permit).
Moreover, if the EPA had responded to Mr. Maier’s petition by promulgating a revised rule, exclusive jurisdiction for review would lie in the Court of Appeals. The fact that the EPA declined to act does not deprive us of jurisdiction, for we have exclusive jurisdiction over “petitions to compel final agency action which would only be reviewable in the United States Courts of Appeal.” See Environmental Defense Fund v. NRC, 902 F.2d 785, 786 (10th Cir.1990).9 This rule ensures that an appellate court will review the Administrator’s decision whether the ultimate challenge is to a failure to revise or to a decision to revise.10 Because exclusive jurisdiction to review the substance of regulations finally promulgated by the EPA lies with the Courts of Appeals, we have jurisdiction to compel revisory rulemaking unless the agency’s failure falls within that class of nondiscretionary duties for which jurisdiction has been granted to the district court.
This is not a ease which could have been brought in district court as a citizen’s suit under section 1365. Such a suit may lie only for failure to perform a nondiscretionary duty. 33 U.S.C. § 1365. The instant case is unlike others in which circuit courts have declined to find section 1369 jurisdiction in the face of the EPA’s refusal to promulgate *1039regulations at all, or its failure to do so by a date certain set by law. For example, in Trustees for Alaska the Ninth Circuit considered a claim that the EPA had totally failed to promulgate specific regulations for the placer mining industry notwithstanding a requirement from Congress to do so, and was instead setting effluent limitations in individual NPDES permits. Trustees for Alaska, 749 F.2d at 558. The court determined that this claim was “framed in terms of the EPA’s failure to comply with a nondiscretionary duty to promulgate industry-wide rules,” id, and was therefore required to be brought in district court under section 1365, id. at 558-59. See also Armco, Inc. v. EPA, 869 F.2d 975, 981-82 (6th Cir.1989) (disclaiming jurisdiction where EPA had refused to perform nondiscretionary responsibility to propose comprehensive sludge management regulations); Pennsylvania Dep’t of Envtl. Resources v. EPA, 618 F.2d 991, 995 (3d Cir.1980) (declining section 1369 jurisdiction over suits to compel EPA to perform nondiscretionary duty to promulgate new source performance standards applicable to post-mining discharges). In these cases, the EPA had failed to issue the disputed regulations at all. In the instant case, the EPA has both issued and revised its regulations, but Mr. Maier contends it has abused its discretion by declining to initiate rulemaking again in light of new information. Both parties agree that the timing of any revision is discretionary.11 Mr. Maier does not contend the EPA has failed to comply with a nondiscretionary duty, and thus a district court could not exercise jurisdiction over his claim under section 1365.12 We therefore hold that we have jurisdiction over the instant case under section 1369.
III.
A.
We review an agency’s refusal to initiate revisory rulemaking to determine if the agency’s refusal was “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A); see Oklahoma v. EPA, 908 F.2d 595, 598 (10th Cir.1990), rev’d on other grounds sub nom. Arkansas v. Oklahoma, 503 U.S. 91, 112 S.Ct. 1046, 117 L.Ed.2d 239 (1992). Although our inquiry into the basis of the agency’s action will be searching and careful, our review is ultimately a narrow one. See Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 823-24, 28 L.Ed.2d 136 (1971). Review under the “arbitrary and capricious” standard “encompasses a range of levels of deference to the agency.” American Horse Protection Ass’n v. Lyng, 812 F.2d 1, 4 (D.C.Cir.1987) (AHPA) (citing WWHT, Inc. v. FCC, 656 F.2d 807, 817 (D.C.Cir.1981)); accord Brown v. Secretary of Health and Human Servs., 46 F.3d 102, 110 (1st Cir.1995). In determining the appropriate level of deference, we heed the nature and context of the challenged agency action or inaction.13
Substantial prudential concerns counsel particularly broad deference in the context of review of an agency refusal to initiate rulemaking. The D.C. Circuit has repeatedly observed that, within the range of deference embodied in the “arbitrary and *1040capricious” standard, refusals to initiate rulemaking are at the high end. See, e.g., Capital Network Sys., Inc. v. FCC, 3 F.3d 1526, 1530 (D.C.Cir.1993) (quoting AHPA 812 F.2d at 4-5; citing Cellnet Communication, Inc. v. FCC, 965 F.2d 1106, 1111 (D.C.Cir.1992)). Courts are ill-equipped and poorly situated to address important reasons for agency inaction, such as the decision that a “problem is not sufficiently important to justify the allocation of significant scarce resources given the nature of the many other problems the agency is attempting to address.” 1 Kenneth C. Davis & Richard J. Pierce, Administrative Law Treatise § 6.9, at 280 (3d ed. 1994). “A court rarely has enough information to second guess agency decisions premised on this type of reasoning.” Id.; see also Natural Resources Defense Council, Inc. v. SEC, 606 F.2d 1031, 1046 (D.C.Cir.1979) (“An agency’s discretionary decision not to regulate a given activity is inevitably based, in large measure, on factors not inherently susceptible to judicial resolution____”); AHPA 812 F.2d at 4-5 (according heightened deference to refusal to initiate rule-making).
Nonetheless, we will not blindly uphold agency refusals to initiate rulemaking in the face of new information. “[Cjhanges in factual and legal circumstances may impose upon the agency an obligation to reconsider a settled policy or explain its failure to do so.” Bechtel v. FCC, 957 F.2d 873, 881 (D.C.Cir.1992). For example, “a refusal to initiate a rulemaking naturally sets off a special alert when a petition has sought a radical modification of a rule on the basis of a radical change in its factual premises.” AHPA 812 F.2d at 5. Thus, the D.C. Circuit has held “that an agency may be forced by a reviewing court to institute rulemaking proceedings if a significant factual predicate of a prior decision on the subject (either to promulgate or not to promulgate specific rules) has been removed.” WWHT, 656 F.2d at 819 (describing Getter v. FCC, 610 F.2d 973 (D.C.Cir.1979)).
An agency determination may also be vulnerable to challenge if it rests on an insufficient legal predicate. Where the agency’s refusal to initiate rulemaking implicates questions of statutory interpretation, we use the familiar Chevron test. When we review an agency’s interpretation of a statute it administers, we ask two questions. “First, always, is the question whether Congress has directly spoken to the precise question at issue. If the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give effect to the unambiguously expressed intent of Congress.” Chevron, U.S.A, Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781, 81 L.Ed.2d 694 (1984). But “if the statute is silent or ambiguous with respect to the specific issue, the question for the court is whether the agency’s answer is based on a permissible construction of the statute.” Id. at 843, 104 S.Ct. at 2782. If Congress has explicitly or implicitly delegated authority to an agency, “legislative regulations are given controlling weight unless they are arbitrary, capricious, or manifestly contrary to the statute.” Id. at 844, 104 S.Ct. at 2782. “This deference is a product both of an awareness of the practical expertise which an agency normally develops, and of a willingness to accord some measure of flexibility to such an agency as it encounters new and unforeseen problems over time.” International Bhd. of Teamsters v. Daniel 439 U.S. 551, 566 n. 20, 99 S.Ct. 790, 800 n. 20, 58 L.Ed.2d 808 (1979).
B.
Mr. Maier asserts here that the existence of new technology mandates revision of the regulations governing publicly owned treatment works. The EPA does not controvert that NOD can now be controlled effectively. The central question on appeal is whether such control must be accomplished through the EPA’s generally-applicable standards for secondary treatment, or whether the EPA may continue to address the problem on a case-by-case basis through the permit process.14
*1041Mr. Maier argues that the EPA’s refusal to initiate rulemaking is arbitrary and capricious because the development of the new technology has removed both the legal and factual predicate of the EPA’s decision not to set parameters for NOD. Mr. Maier asserts that because the CWA is a “technology-forcing” statute, the development of cost-effective new technology must be incorporated into the EPA’s generally-applieable secondary treatment regulations. Alternatively, he argues that even if the statute gives the EPA discretion to address NOD levels by permit, its decision to do so in this instance is not supported by the evidence before the agency or is based on the consideration of impermissible factors. Athough there is substantial overlap between these arguments, we consider them in turn.
Under Chevron’s first step, we ask “whether Congress has directly spoken to the precise question at issue.” 467 U.S. at 842, 104 S.Ct. at 2781. Section 1311 of the CWA specifies:
(a) Except as in compliance with this [and other] sections ... the discharge of any pollutant by any person shall be unlawful.
(b) In order to carry out the objective of this chapter there shall be achieved—
(1)(B) for publicly owned treatment works ..., effluent limitations based upon secondary treatment as defined by the Administrator pursuant to section 1314(d)(1) of this title____
33 U.S.C. § 1311 (emphasis added). In turn, section 1314(d) provides that “[t]he Administrator ... shall publish within sixty days after October 18, 1972 (and from time to time thereafter) information, in terms of amounts of constituents and chemical, physical, and biological characteristics of pollutants, on the degree of effluent reduction attainable through the application of secondary treatment.” Id. § 1314(d)(1). Aside from sections 1311 and 1314, the CWA does not further delimit “secondary treatment,” or specifically constrain the Administrator in promulgating generally-applieable effluent limitations for POTWs.
Congress has not directly spoken to the precise question of whether the EPA has discretion to conclude that reductions in NOD are not required to be achieved by generally-applieable effluent limitations, but may instead be imposed by permit. The statute plainly delegates to the EPA authority to define secondary treatment, and to promulgate generally-applieable regulations based on its definition. We must therefore determine whether the EPA’s secondary treatment regulations are a “permissible construction” of the CWA Mr. Maier posits that the statutory definition of “secondary treatment” must include technology to reduce NOD and nutrients if such technology is available and cost-effective. He argues that the EPA’s decision to control NOD by permit is not a legitimate policy choice because the CWA requires the EPA to promulgate generally-applicable effluent limitations for every pollutant that can be controlled with “secondary treatment.” We disagree and conclude that the EPA’s secondary treatment regulations are a permissible exercise of its authority under sections 1311 and 1314.
As an initial matter, we must reject Mr. Maier’s claim that § 1314(d)(1) requires the Administrator to publish secondary treatment regulations for any pollutant that can be controlled via secondary treatment. That provision plainly mandates only that the EPA issue scientific information “from time to time.” Compare 33 U.S.C. § 1314(d)(1) (“The Administrator ... shall publish ... from time to time ... information ... on the degree of effluent reduction attainable through the application of secondary treatment.”), with id. § 1314(b) (“the Administrator shall ... publish ... regulations.”). While it is true that where reductions in NOD are “attainable through the application of secondary treatment,” id. § 1314(d)(1), the EPA may have a present duty under § 1314(d)(1) to publish information pertaining to those reductions, cf. Oljato, 515 F.2d at 662, the EPA is not required under that same provision to issue regulations limiting *1042NOD discharges from POTWs. Mr. Maier has not advanced a duty-to-publish claim in the instant case, and we therefore do not consider this question further.
As a matter of statutory delegation and practical necessity, the EPA exercises its expertise to determine if a given technology ought to form the basis of the standard “secondary treatment” defined under section 1314(d)(1). The phrase “secondary treatment” has an independent meaning apart from its statutory context. “Conventional or secondary treatment of municipal waste includes biological processes, primarily decomposition, with or without chemical disinfectants, to remove organic wastes.” City of Sarasota, 813 F.2d at 1108 n. 7. The phrase distinguishes one stage in a variegated treatment system:
There are three levels of wastewater treatment. Primary treatment refers to a physical sedimentation process for removing settleable solids. Secondary treatment refers to a physical/biological process for removing solids and pollutants characterized by biological oxygen demand and pH. Tertiary treatment involves processes which remove other pollutants such as non-biodegradable toxics.
Natural Resources Defense Council, Inc. v. EPA 790 F.2d at 293 n. 2; see also California v. EPA 689 F.2d 217, 218 (D.C.Cir.1982) (distinguishing secondary treatment, advance secondary treatment, and advance waste treatment). Although these descriptions suggest that NOD and nutrients fall within a general understanding of secondary treatment, they also demonstrate “secondary treatment” has a broad connotation.
The legislative history of the CWA also offers little guidance to the statutory definition of “secondary treatment” with respect to NOD. The EPA argues Congress intended that “secondary treatment” concern “organic” oxygen depletion, i.e., CBOD, but not NOD and nutrients. However, it is far from clear from the legislative history that this distinction was ever considered. As noted in the House Report, “[secondary treatment as considered in the context of a publicly owned treatment works is generally concerned with suspended solids and biologically degradable, oxygen demanding materials (BOD).” H. Rep. No. 92-911, at 101 (1972). It appears more likely that when the CWA was enacted, treatment for NOD simply was not technologically feasible. See, e.g., Secondary Treatment Regulation, 49 Fed.Reg. 36,986, 36,988 (1984) (to be codified at 40 C.F.R. pt. 133) (“Secondary treatment requirements are based on controlling the oxygen demand due to the carbonaceous component of the organic material in the effluent because secondary treatment facilities can effectively remove carbonaceous organic material ... but may not consistently remove ammonia.”); Secondary Treatment Information, 48 Fed.Reg. 52,-272, 52,273 (1983) (“‘[NJutrients ... were not specified for inclusion, because secondary treatment, under normal conditions, does not effectively or consistently remove them.’”) (citation omitted). Congress did not speak to the problem confronting us because for practical purposes it did not yet exist.
Given the broad category of treatment processes and technology encompassed by “secondary treatment,” and statutory delegation to the Administrator to define secondary treatment, we do not find it impermissible for the EPA to refuse to extend its definition of standard secondary treatment to include controls on NOD. The EPA has consistently classified NOD reduction as a form of “advanced treatment” that will be required by permit if necessary to protect water quality. See, e.g., Secondary Treatment Information, 48 Fed.Reg. 52,272, 52,275 (1983) (proposed Nov. 16, 1983) (describing nitrification processes as “beyond secondary” treatment). We do not agree with Mr. Maier or the dissent that technological feasibility is the only criterion the EPA may use to determine which of the universe of secondary treatment technologies ought to be considered standard. Here, the agency’s choice is supported by its reasoned consideration of other factors that lie within its expertise in administering the statute.15
*1043Section 1311(b)(1)(B), in conjunction with section 1314(d)(1), provides for the promulgation of generally-applicable effluent limitations for POTWs. Section 1311(b)(1)(B) gives the EPA authority to determine the stringency and scope of generally-applicable effluent limitations that are based on secondary treatment. Thus, even if reductions of NOD and nutrients potentially fall within the definition of “secondary treatment,” the EPA must determine if it should promulgate generally-applicable effluent limitations for these specific pollutants. The statute requires that generally-applicable effluent limitations for POTWs be “based upon secondary treatment.” 33 U.S.C. § 1311(b)(1)(B) (emphasis added). Contrary to Mr. Maier’s assertion, the statute does not on its face require that the generally-applicable effluent limitations address all pollutants that might be reduced by secondary treatment.16 When we apply the deference due an administrative agency which Chevron mandates, “[t]he permissive nature of the statute implies broad agency discretion in selecting the appropriate manner of regulation.” Professional Drivers Council v. Bureau of Motor Carrier Safety, 706 F.2d 1216, 1221 (D.C.Cir.1983). EPA’s position that the statute allows the agency to impose limits for NOD on a case-by-case basis through the permitting process is a reasonable and permissible reading of the statute, to which we must defer.17
The EPA’s exercise of its discretion under sections 1311 and 1314 is not in these circumstances arbitrary, capricious, or manifestly contrary to the statute. The EPA has made the uncontroverted claim that the effect of NOD is highly variable and site-specific, and thus not appropriate to a general regulation applicable to every POTW. Here, the EPA and the States approved to administer the NPDES permit program routinely impose NOD and nutrient limitations on POTWs on a case-by-case basis by permit. We are not faced with a situation in which the EPA has chosen to ignore a pollutant or category of pollutants for which effluent reductions are attainable by secondary treatment. The EPA has not substituted control by permit for control by generally-applicable effluent limitation without a reasoned explanation for its choice of method. Contrary to the dissent’s suggestion, the EPA has not sought, nor do we approve, “general discretion to define secondary treatment to cover only those pollutants that are — in the view of the Administrator — more appropriately regulated via generally-applicable regulations rather than ease-by-case quality-based limits.” Dissent at 1049. The EPA’s discretion is not unbridled.
This brings us to Mr. Maier’s second argument, which is that the EPA’s refusal to include parameters for NOD and ultimate BOD in its secondary treatment regulations in this instance was nonetheless arbitrary and capricious because the refusal was not supported by the evidence and was based on a consideration of impermissible factors. The Supreme Court has stated
*1044an agency rule would be arbitrary and capricious if the agency has relied on factors which Congress has not intended it to consider, entirely failed to consider an important aspect of the problem, offered an explanation for its decision that runs counter to the evidence before the agency, or is so implausible that it could not be ascribed to a difference in view or the product of agency expertise.
Motor Vehicle Mfrs. Ass’n v. State Farm, Mutual Auto. Ins. Co., 463 U.S. 29, 43, 103 S.Ct. 2856, 2867, 77 L.Ed.2d 443 (1983). The Court’s statement is equally applicable to an agency’s basis for declining to make a rule.
The EPA does not dispute that one factual predicate of the existing regulations has changed — the feasibility and cost-effectiveness of technology to control NOD. See WWHT, 656 F.2d at 819 (“[A]n agency may be forced by a reviewing court to institute rulemaking proceedings if a significant factual predicate of a prior decision on the subject ... has been removed.”). Mr. Maier argues that technological feasibility is the only criterion the EPA may use to determine which secondary treatment controls ought to be generally applicable, and which may be imposed on a case-by-case basis. We disagree. The agency’s decision to control NOD and nutrients by permit is supported by two other factual predicates that lie within its expertise in administering the statute: first, the impact of NOD and nutrients on water quality is highly variable with the characteristics of the receiving body of water; second, control of NOD by permit adequately protects water quality where necessary. Mr. Maier does not dispute these factual premises. Instead, he argues that the first is a legally impermissible consideration, and he explicitly eschews a challenge to the second.18
Mr. Maier observes Congress intended the effluent limitations in the CWA to be technology-based, not based on assessments of water quality. In general, we agree. See EPA v. California ex rel. State Water Resources Control Board, 426 U.S. 200, 204-05, 96 S.Ct. 2022, 2024-25, 48 L.Ed.2d 578 (1976). However, effluent limitations for POTWs must be “based upon” secondary treatment technology, not eo-extensive with it. We do not agree with Mr. Maier that it is impermissible for the EPA to consider effects on water quality in determining whether reductions attainable by new secondary treatment technology ought to be uniformly imposed on all POTWs. The purpose of the CWA is to “restore and maintain the chemical, physical, and biological integrity of the Nation’s waters....” 33 U.S.C. § 1251(a). The EPA relies on statutory language authorizing the EPA to use the permitting process to impose “more stringent limitation[s]” on POTWs where necessary to protect water quality. 33 U.S.C. § 1311(b)(1)(C). We need not deter- ■ mine if section 1311(b)(1)(C) directly authorizes the use of permitting as an alternative to, as opposed to a supplement for, generally-applicable effluent limitations.19 The fact that Congress has, in this closely related statutory section, provided for quality-based permitting as a gap-filling measure gives strong support to the EPA’s exercise of delegated authority to fill the gap where it has concluded that NOD should not be part of standard secondary treatment. The fact that secondary treatment controls are technology-based does not preclude the EPA from deciding that certain technologically-attainable standards are necessary and appropriate only for some POTWs. We should not order the agency to develop generally-applicable parameters based on the use of new technology, even if cost-effective, in the face of the agency’s reasoned judgment that the use of such technology is irrelevant to the attainment of water quality standards in many circumstances.
Contrary to the suggestion of the dissent, we do not approve the EPA’s regulations based on the agency’s naked “policy preference for quality-based controls rather than *1045generally applicable limitations.” Dissent at 1048. The EPA has articulated its uncontroverted view that NOD is highly variable with the conditions of the receiving body of water. Consequently, NOD is particularly unsuited for a generally applicable regulation, and is appropriately dealt with — and is being dealt with — by permit. We agree with the dissent that the Clean Water Act amendments created a statutory regime in which technology-based standards are the primary mechanism of controlling discharge of pollutants. It is a far stretch from this presumption to the conclusion that the EPA may not exercise its authority in filling the gaps to conclude that certain pollutants, due to their peculiar characteristics, need not be controlled by across-the-board standards. Contrary to the dissent’s inference, we do not hold that the EPA may choose a permit-based approach over a technology-based standard merely based on the agency’s “policy preference.”
The dissent gives lip service to the notion of deference to the agency, but the deference it would give is cramped indeed. The dissent asserts that to satisfy Chevron, “the implementing agency must point to some language in the statute to justify its policy conclusion — here, that the POTW regulatory regime can legitimately depart from the core public policy of the Clean Water Act.” Dissent at 1048-49. This turns the Chevron test on its head: the dissent, rather than deferring to the agency’s gap-filling unless “manifestly contrary to the statute,” imposes on the agency the burden of proving that its gap-filling is manifestly authorized by the statute — in which case it would be unnecessary to invoke the Chevron framework in the first instance. We are faced with a paradigmatic situation calling for deference to the agency: where the statute allows the agency to exercise discretion, we should defer to the agency’s use of its expertise in striking a careful balance between the broad statutory purposes and the unique problems posed by specific pollutants and technologies. The EPA has offered a reasoned basis for its belief that this balance is to apply a technology-based standard to some, rather than all, POTWs.
Mr. Maier argues that, even if the EPA is right about the costs of retrofitting existing POTWs, new facilities can be designed to treat NOD at no greater expense and the EPA should so require. Even if this is true, we are not persuaded the EPA’s failure to so act is arbitrary or capricious. The EPA maintains that NOD is adequately addressed on a plant-by-plant basis. Mr. Maier has failed to demonstrate that stricter standards in individual permits are not effective in promoting the building of these newer, more effective designs. Moreover, this argument ignores the real costs of administering environmental laws. Promulgating revised regulations necessitates a substantial commitment of limited agency resources. We have “little ability to determine the resources available to the agency or to determine whether the other problems to which the agency has chosen to devote its scarce resources are more or less important than the problem raised in the petition.” 1 Davis & Pierce, Administrative Law Treatise § 6.9, at 280.
On the record in this ease, we cannot hold that the EPA’s interpretation of the CWA is arbitrary or capricious, or that its rejection of rulemaking in this instance is arbitrary or capricious. Since the EPA first promulgated regulations under Section 1311, it has never required that all oxygen-depleting pollutants be removed by means of generally-applicable secondary treatment controls. These regulations do not stand alone; rather, they set a national floor for the performance of secondary treatment systems. Primary and tertiary treatments complement their function, and individual permits for treatment facilities can have stricter standards. EPA makes a reasoned argument that where NOD is a problem, it may be addressed in the terms of a POTWs permit, and points out that 53% of major secondary treatment facilities across the country now have such requirements. Ree. at 123. Mr. Maier would have to provide impressive evidence that the EPA has acted irrationally. He has not done so. Without a more convincing showing that the nation’s municipal water treatment system is broken, we will not order the EPA to fix it.
AFFIRMED.
. The Utah Wilderness Association was dismissed from this appeal. For simplicity’s sake, we shall refer only to Mr. Maier as petitioner.
. Secondary treatment generally refers to a process of physical and biological treatment of wastewater to remove pollutants which deplete the water's oxygen content and increase its acidity. See generally City of Sarasota v. EPA, 813 F.2d 1106, 1108 n. 7 (11th Cir.1987); Natural Resources Defense Council, Inc. v. EPA, 790 F.2d 289, 293 n. 2 (3d Cir.1986).
. Carbonaceous biochemical oxygen demand (CBOD) quantifies "the amount of oxygen consumed by various microorganisms in metabolizing organic (carbon) matter in the wastewater,” while nitrogenous biochemical oxygen demand (NOD) measures “the oxygen consumed by other types of bacteria ... in converting ammonia to nitrite and then to nitrate, a process known as nitrification.” Secondary Treatment Information, 48 Fed.Reg. 52,272, 52,274 (1983) (proposed Nov. 16, 1983). NOD can also be referred to as NBOD, as Mr. Maier does.
“Nutrients" are related to NOD, and refer to materials upon which certain bacteria feed. Oxygen is consumed when these bacteria respire, or when they die and other organisms use oxygen to consume them.
. Controlling CBOD has been the main goal of secondary municipal wastewater treatment both because CBOD is usually a much greater problem than NOD and because the development of technology to effectively address high levels of NOD has lagged behind that used to reduce CBOD. Secondary Treatment Information, 48 Fed.Reg. 52,272, 52,275 (1983) (proposed Nov. 16, 1983).
More than a decade ago, scientific commentators speculated as to whether NOD was meant to be controlled as a component of BOD. See rec. at 43 (“Although significant nitrification can occur in the BOD5 test for secondary effluents, much debate centers on whether the 30-mg/L BOD5 standard, as defined by EPA, was intended to include only carbonaceous oxidation, or also any nitrogenous oxidation that might be exerted in the BOD5 test.”).
. Standard secondary treatment did not include controls on NOD.
. CBOD levels tend to rise sharply and then plateau, while NOD levels increase at a slower rate. As a result, early testing for BOD (e.g., after five days) will accurately assess CBOD but will fail to measure the levels to which NOD will eventually rise. This rise in NOD would be captured by the measurement of ultimate BOD. This disparity has apparently been recognized for several decades, see rec. at 21, 51, but Mr. Maier suggests it was ignored when EPA promulgated regulations under the CWA, see id. at 8; see also id. at 244.
. Section 1369 codifies Section 509 of the Feder- of 1972. al Water Pollution Control Act (Clean Water Act)
.We note that two courts have eschewed jurisdiction under section 1369; however, neither case involved the continuing validity of a rule in the face of new information. In Bethlehem Steel Corp. v. EPA, 782 F.2d 645, 656 (7th Cir.1986), the remedy sought was "not to rescind or modify” as a result of new information, but rather "to conduct a new, follow-on rulemaking proceeding ... distinct from that which [was] under review.” Likewise, in Natural Resources Defense Council, Inc. v. Train, 519 F.2d 287, 291 (D.C.Cir.1975), review was sought for "the omission of certain substances” and ”[u]nless a substance is listed no standard or prohibition reviewable under section [1369] will ever be promulgated.” We also note that review was not possible because neither case had a complete administrative record. Bethlehem Steel, 782 F.2d at 655-56 (review "will require the compilation of a new administrative record”); Natural Resources Defense Council, Inc. v. Train, 519 F.2d at 291 ("[T]he Administrator had not filed the entire administrative record with the court.").
. Although the statutory scheme in Environmental Defense Fund v. NRC did not contain a bifurcated system of judicial review, that fact is irrelevant. The CWA confers jurisdiction on the district courts only for the failure to perform non-discretionaiy duties. Where the alleged duty is discretionary, as here, the courts of appeals have jurisdiction.
. Prudential concerns are not to the contrary, for “[u]nlike other actions in which the district court serves the important functions of developing a complete record and ‘defining and focusing the issues,' in many appeals of administrative actions, the agency already has fulfilled those functions and there is no practical reason to submit the issues to two-tiered judicial review.” 2 Steven Alan Childress & Martha S. Davis, Federal Standards of Review § 14.03, at 14-19 (2d ed. 1992) (citation omitted).
. We note the dicta in Oljato Chapter that a duly under the CAA to revise regulations “from time to time” is discretionary was rendered obsolete by statutory revision imposing a mandatory duty to review regulations every five years. See Environmental Defense Fund v. Thomas, 870 F.2d 892, 896-97 (2d Cir.1989).
. As the court in Oljato Chapter noted, the limited jurisdiction granted to the district court would be rendered boundless if an abuse of discretion were considered to be a "failure to perform a nondiscretionary act.” 515 F.2d at 662 (emphasis added).
.In Heckler v. Chaney, 470 U.S. 821, 831, 105 S.Ct. 1649, 1655-56, 84 L.Ed.2d 714 (1985), the Supreme Court decided that agency refusals to take specific enforcement actions are presumptively unreviewable under the APA. However, the Court expressly did not address the level of deference due when the agency refuses to initiate rulemaking. Id. at 825 n. 2, 105 S.Ct. at 1652 n. 2. The D.C. Circuit has held that the Chaney presumption does not apply to refusals to initiate rulemaking. AHPA, 812 F.2d at 4-5. See also Cass R. Sunstein, Reviewing Agency Inaction After Heckler v. Chaney, 52 U.ChiL.Rev. 653, 680-83 (1985). The EPA agrees that its refusal to initiate rulemaking is reviewable.
. In his initial appellate brief, Mr. Maier also argued that the EPA must amend its regulations to replace testing for oxygen depletion after five days (BOD5) with testing for ultimate oxygen *1041depletion. In his reply brief, Mr. Maier concedes that this argument cannot stand alone and is founded on his argument about the need to set general parameters for NOD.
. The dissent claims the EPA is foreclosed from exercising its discretion to exclude NOD from its definition of secondary treatment because the agency has used secondary treatment regulations to control BOD, of which NOD is a component. As our discussion in part I.B makes clear, how*1043ever, BOD parameters have historically been set by the agency as a rough proxy for CBOD. At the time that the BOD limitations were first promulgated, control of the NOD component was not technologically feasible. As the dissent constantly reminds us, secondary treatment is technology-based. The BOD parameter was thus targeted at CBOD, for which reductions were technologically attainable, and not at NOD, for which reductions were not technologically attainable. By the time of the EPA’s 1984 rulemaking, a nitrification-inhibited test was developed which allowed for the isolation of the CBOD component. The EPA promulgated regulations approving the use of the CBOD test in lieu of the general BOD test. The EPA has never set discrete limitations on the NOD component.
. On its face, the CWA gives the EPA less discretion in setting effluent limitations on point sources other than POTWs. See 33 U.S.C. § 1311(b)(2)(A) (stating that for certain identified toxic pollutants “there shall be achieved ... effluent limitations ... which shall require the application of the best available technology economically achievable.”).
. The dissent suggests that our analysis of section 1311 substitutes our own construction of the statute for that advanced by the agency. In so doing, the dissent fails to distinguish between the source of agency discretion, which we must determine in the first instance under Chevron, and the basis for the agency’s exercise of its discretion, for which the dissent correctly observes we may not supply our own rationale. Moreover, we may not avoid construing the statute to the extent necessary to determine that the agency’s regulations are not "manifestly contrary to the statute.” Chevron, 467 U.S. at 844, 104 S.Ct. at 2782.
. In his Reply Brief, Mr. Maier appends material pertaining to the effectiveness of the permitting process in protecting water quality from NOD. We decline to consider this material as it was not presented in the petition to the EPA for consideration by the agency.
. The dissent asserts that the "more stringent” permits described in section 1311(b)(1)(C) may only come into play where a generally-applicable regulation is not technologically feasible, or as a supplement when the generally-applicable regulation is inadequate to protect water quality.