delivered the opinion of the Court.
These cases present the following questions: (1) Whether § 109(b)(1) of the Clean Air Act (CAA) delegates legislative power to the Administrator of the Environmental Protection Agency (EPA). (2) Whether the Administrator may consider the costs of implementation in setting national ambient air quality standards (NAAQS) under § 109(b)(1). (3) Whether the Court of Appeals had jurisdiction to review the EPA’s interpretation of Part D of Title I of the CAA, 42 U. S. C. §§ 7501-7515, with respect to implementing the revised ozone NAAQS. (4) If so, whether the EPA’s interpretation of that part was permissible.
I
. Section 109(a) of the CAA, as added, 84 Stat. 1679, and amended, 42 U. S. C. § 7409(a), requires the Administrator of the EPA to promulgate NAAQS for each air pollutant for which “air quality criteria” have been issued under § 108, 42 U. S. C. § 7408. Once a NAAQS has been promulgated, the Administrator must review the standard (and the criteria *463on which it is based) “at five-year intervals” and make “such revisions ... as may be appropriate.” CAA § 109(d)(1), 42 U. S. C. § 7409(d)(1). These cases arose when, on July 18, 1997, the Administrator revised the NAAQS for particulate matter and ozone. See NAAQS for Particulate Matter, 62 Fed. Reg. 38652 (codified in 40 CFR §50.7 (1999)); NAAQS for Ozone, id., at 38856 (codified in 40 CFR §§50.9, 50.10 (1999)). American Trucking Associations, Inc., and its corespondents in No. 99-1257 — which include, in addition to other private companies, the States of Michigan, Ohio, and West Virginia — challenged the new standards in the Court of Appeals for the District of Columbia Circuit, pursuant to 42 U. S. C. § 7607(b)(1).
The District of Columbia Circuit accepted some of the challenges and rejected others. It agreed with the No. 99-1257 respondents (hereinafter respondents) that § 109(b)(1) delegated legislative power to the Administrator in contravention of the United States Constitution, Art. I, § 1, because it found that the EPA had interpreted the statute to provide no “intelligible principle” to guide the agency’s exercise of authority. American Trucking Assns., Inc. v. EPA, 175 F. 3d 1027, 1034 (1999). The court thought, however, that the EPA could perhaps avoid the unconstitutional delegation by adopting a restrictive construction of § 109(b)(1), so instead of declaring the section unconstitutional the court remanded the NAAQS to the agency. Id., at 1038. (On this delegation point, Judge Tatel dissented, finding the statute constitutional as written. Id., at 1057.) On the second issue that the Court of Appeals addressed, it unanimously rejected respondents’ argument that the court should depart from the rule of Lead Industries Assn., Inc. v. EPA, 647 F. 2d 1130, 1148 (CADC 1980), that the EPA may not consider the cost of implementing a NAAQS in setting the initial standard. It also rejected respondents’ argument that the implementation provisions for ozone found in Part D, Sub-part 2, of Title I of the CAA, 42 U. S. C. §§ 7511-7511f, were *464so tied to the existing ozone standard that the EPA lacked the power to revise the standard. The court held that although Subpart 2 constrained the agency’s method of implementing the new standard, 175 F. 3d, at 1050, it did not prevent the EPA from revising the standard and designating areas of the country as “nonattainment areas,” see 42 U. S. C. § 7407(d)(1), by reference to it, 175 F. 3d, at 1047-1048. On the EPA’s petition for rehearing, the panel adhered to its position on these points, and unanimously rejected the EPA’s new argument that the court lacked jurisdiction to reach the implementation question because there had been no “final” implementation action. American Trucking Assns., Inc. v. EPA, 195 F. 3d 4 (CADC 1999). The Court of Appeals denied the EPA’s suggestion for rehearing en banc, with five judges dissenting. Id., at 13.
The Administrator and the EPA petitioned this Court for review of the first, third, and fourth questions described in the first paragraph of this opinion. Respondents conditionally cross-petitioned for review of the second question. We granted certiorari on both petitions, 529 U. S. 1129 (2000); 530 U. S. 1202 (2000), and scheduled the cases for argument in tandem. We have now consolidated the cases for purposes of decision.
II
In Lead Industries Assn., Inc. v. EPA, supra, at 1148, the District of Columbia Circuit held that “economic considerations [may] play no part in the promulgation of ambient air quality standards under Section 109” of the CAA. In the present cases, the court adhered to that holding, 175 F. 3d, at 1040-1041, as it had done on many other occasions. See, e. g., American Lung Assn. v. EPA, 134 F. 3d 388, 389 (1998); NRDC v. Administrator, EPA, 902 F. 2d 962, 973 (1990), vacated in part on other grounds, NRDC v. EPA, 921 F. 2d 326 (CADC 1991); American Petroleum Institute v. Costle, 665 F. 2d 1176, 1185 (1981). Respondents argue that these *465decisions are incorrect. We disagree; and since the first step in assessing whether a statute delegates legislative power is to determine what authority the statute confers, we address that issue of interpretation first and reach respondents’ constitutional arguments in Part III, infra.
Section 109(b)(1) instructs the EPA to set primary ambient air quality standards “the attainment and maintenance of which ... are requisite to protect the public health” with “an adequate margin of safety.” 42 U. S. C. § 7409(b)(1). Were it not for the hundreds of pages of briefing respondents have submitted on the issue, one would have thought it fairly clear that this text does not permit the EPA to consider costs in setting the standards. The language, as one scholar has noted, “is absolute.” D. Currie, Air Pollution: Federal Law and Analysis 4-15 (1981). The EPA, “based on” the information about health effects contained in the technical “criteria” documents compiled under § 108(a)(2), 42 U. S. C. § 7408(a)(2), is to identify the maximum airborne concentration of a pollutant that the public health can tolerate, decrease the concentration to provide an “adequate” margin of safety, and set the standard at that level. Nowhere are the costs of achieving such a standard made part of that initial calculation.
Against this most natural of readings, respondents make a lengthy, spirited, but ultimately unsuccessful attack. They begin with the object of § 109(b)(l)’s focus, the “public health.” When the term first appeared in federal clean air legislation — in the Act of July 14, 1955 (1955 Act), 69 Stat. 322, which expressed “recognition of the dangers to the public health” from air pollution — its ordinary meaning was “[t]he health of the community.” Webster’s New International Dictionary 2005 (2d ed. 1950). Respondents argue, however, that § 109(b)(1), as added by the Clean Air Amendments of 1970,84 Stat. 1676, meant to use the term’s secondary meaning: “[t]he ways and means of conserving the health *466of the members of a community, as by preventive medicine, organized care of the sick, etc.” Ibid. Words that can have more than one meaning are given content, however, by their surroundings, FDA v. Brown & Williamson Tobacco Corp., 529 U. S. 120, 182-183 (2000); Jones v. United States, 527 U. S. 373, 389 (1999), and in the context of § 109(b)(1) this second definition makes no sense. Congress could not have meant to instruct the Administrator to set NAAQS at a level “requisite to protect” “the art and science dealing with the protection and improvement of community health.” Webster’s Third New International Dictionary 1836 (1981). We therefore revert to the primary definition of the term: the health of the public.
Even so, respondents argue, many more factors than air pollution affect public health. In particular, the economic cost of implementing a very stringent standard might produce health losses sufficient to offset the health gains achieved in cleaning the air — for example, by closing down whole industries and thereby impoverishing the workers and consumers dependent upon those industries. That is unquestionably true, and Congress was unquestionably aware of it. Thus, Congress had commissioned in the Air Quality Act of 1967 (1967 Act) “a detailed estimate of the cost of carrying out the provisions of this Act; a comprehensive study of the cost of program implementation by affected units of government; and a comprehensive study of the economic impact of air quality standards on the Nations industries, communities, and other contributing sources of pollution.” §2, 81 Stat. 505. The 1970 Congress, armed with the results of this study, see The Cost of Clean Air, S. Doc. No. 91-40 (1969) (publishing the results of the study), not only anticipated that compliance costs could injure the public health, but provided for that precise exigency. Section 110(f)(1) of the CAA permitted the Administrator to waive the compliance deadline for stationary sources if, inter *467alia, sufficient control measures were simply unavailable and “the continued operation of such sources is essential... to the public health or welfare.” 84 Stat. 1683 (emphasis added). Other provisions explicitly permitted or required economic costs to be taken into account in implementing the air quality standards. Section 111(b)(1)(B), for example, commanded the Administrator to set “standards of performance” for certain new sources of emissions that as specified in § 111(a)(1) were to “reflec[t] the degree of emission limitation achievable through the application of the best system of emission reduction which (taking into account the cost of achieving such reduction) the Administrator determines has been adequately demonstrated.” Section 202(a)(2) prescribed that emissions standards for automobiles could take effect only “after such period as the Administrator finds necessary to permit the development and application of the requisite technology, giving appropriate consideration to the cost of compliance within such period.” 84 Stat. 1690. See also § 202(b)(5)(C) (similar limitation for interim standards); § 211(c)(2) (similar limitation for fuel additives); § 231(b) (similar limitation for implementation of aircraft emission standards). Subsequent amendments to the CAA have added many more provisions directing, in explicit language, that the Administrator consider costs in performing various duties. See, e. g., 42 U. S. C. § 7545(k)(1) (reformulate gasoline to “require the greatest reduction in emissions ... taking into consideration the cost of achieving such emissions reductions”); § 7547(a)(3) (emission reduction for nonroad vehicles to be set “giving appropriate consideration to the cost” of the standards). We have therefore refused to find implicit in ambiguous sections of the CAA an authorization to consider costs that has elsewhere, and so often, been expressly granted. See Union Elec. Co. v. EPA, 427 U. S. 246, 257, and n. 5 (1976). Cf. General Motors Corp. v. United States, 496 U. S. 530, 538, 541 (1990) *468(refusing to infer in certain provisions of the CAA deadlines and enforcement limitations that had been expressly imposed elsewhere).
Accordingly, to prevail in their present challenge, respondents must show a textual commitment of authority to the EPA to consider costs in setting NAAQS under § 109(b)(1). And because § 109(b)(1) and the NAAQS for which it provides are the engine that drives nearly all of Title I of the CAA, 42 U. S. C. §§7401-7515, that textual commitment must be a clear one. Congress, we have held, does not alter the fundamental details of a regulatory scheme in vague terms or ancillary provisions — it does not, one might say, hide elephants in mouseholes. See MCI Telecommunications Corp. v. American Telephone & Telegraph Co., 512 U. S. 218, 231 (1994); FDA v. Brown & Williamson Tobacco Corp., supra, at 159-160. Respondents’ textual arguments ultimately founder upon this principle.
Their first claim is that §109(b)(l)’s terms “adequate margin” and “requisite” leave room to pad health effects with cost concerns. Just as we found it “highly unlikely that Congress would leave the determination of whether an industry will be entirely, or even substantially, rate-regulated to agency discretion — and even more unlikely that it would achieve that through such a subtle device as permission to ‘modify’ rate-filing requirements,” MCI Telecommunications Corp. v. American Telephone & Telegraph Co., supra, at 231, so also we find it implausible that Congress would give to the EPA through these modest words the power to determine whether implementation costs should moderate national air quality standards. Accord, Christensen v. Harris County, 529 U. S. 576, 590, n. (2000) (Scalia, J., concurring in part and concurring in judgment) (“The implausibility of Congress’s leaving a highly significant issue unaddressed (and thus ‘delegating’ its resolution to the administering agency) is assuredly one of the factors *469to be considered in determining whether there is ambiguity” (emphasis deleted)).1
The same defect inheres in respondents’ next two arguments: that while the Administrator’s judgment about what is requisite to protect the public health must be “based on [the] criteria” documents developed under § 108(a)(2), see § 109(b)(1), it need not be based solely on those criteria; and that those criteria themselves, while they must include “effects on public health or welfare which may be expected from the presence of such pollutant in the ambient air,” are not necessarily limited to those effects. Even if we were to concede those premises, we still would not conclude' that one of the unenumerated factors that the agency can consider in developing and applying the criteria is cost of implementation. That factor is both so indirectly related to public health and so full of potential for canceling the conclusions drawn from direct health effects that it would surely have been expressly mentioned in §§ 108 and 109 had Congress meant it to be considered. Yet while those provisions describe in detail how the health effects of pollutants in the ambient air are to be calculated and given effect, see § 108(a)(2), they say not a word about costs.
Respondents point, finally, to a number of provisions in the CAA that do require attainment cost data to be generated. Section 108(b)(1), for example, instructs the Administrator to “issue to the States,” simultaneously with the criteria documents, “information on air pollution control techniques, which information shall include data relating to the cost of installation and operation.” 42 U. S. C. § 7408(b)(1). And *470§ 109(d)(2)(C)(iv) requires the Clean Air Scientific Advisory Committee to “advise the Administrator of any adverse public health, welfare, social, economic, or energy effects which may result from various strategies for attainment and maintenance” of NAAQS.2 42 U. S. C. § 7409(d)(2)(C)(iv). Respondents argue that these provisions make no sense unless costs are to be considered in setting the NAAQS. That is not so. These provisions enable the Administrator to assist the States in carrying out their statutory role as primary implementers of the NAAQS. It is to the States that the CAA assigns initial and primary responsibility for deciding what emissions reductions will be required from which sources. See 42 U. S. C. §§ 7407(a), 7410 (giving States the duty of developing implementation plans). It would be impossible to perform that task intelligently without considering which abatement technologies are most efficient, and most economically feasible — which is why we have said that “the most important forum for consideration of claims of economic and technological infeasibility is before the state agency formulating the implementation plan,” Union Elec. Co. v. EPA, 427 U. S., at 266. Thus, federal clean air legislation has, from the very beginning, directed federal agencies to develop and transmit implementation data, including cost data, to the States. See 1955 Act, *471§2(b), 69 Stat. 322; Clean Air Act of 1963, amending §§3(a), (b) of the CAA, 77 Stat. 394; 1967 Act, §§ 103(a)-(d), 104, 107(c), 81 Stat. 486-488. That Congress chose to carry forward this research program to assist States in choosing the means through which they would implement the standards is perfectly sensible, and has no bearing upon whether cost considerations are to be taken into account in formulating the standards.3
It should be clear from what we have said that the canon requiring texts to be so construed as to avoid serious constitutional problems has no application here. No matter how severe the constitutional doubt, courts may choose only between reasonably available interpretations of a text. See, e. g., Miller v. French, 530 U. S. 327, 341 (2000); Pennsylvania Dept. of Corrections v. Yeskey, 524 U. S. 206, 212 (1998). The text of § 109(b), interpreted in its statutory and historical context and with appreciation for its importance to the CAA as a whole, unambiguously bars cost considerations from the NAAQS-setting process, and thus ends the matter for us as well as the EPA.4 We therefore affirm the judgment of the Court of Appeals on this point.
*472III
Section 109(b)(1) of the CAA instructs the EPA to set “ambient air quality standards the attainment and maintenance of which in the judgment of the Administrator, based on [the] criteria [documents of § 108] and allowing an adequate margin of safety, are requisite to protect the public health.” 42 U. S. C. § 7409(b)(1). The Court of Appeals held that this section as interpreted by the Administrator did not provide an “intelligible principle” to guide the EPA’s exercise of authority in setting NAAQS. “[The] EPA,” it said, “lack[ed] any determinate criteria for drawing lines. It has failed to state intelligibly how much is too much.” 175 F. 3d, at 1034. The court hence found that the EPA’s interpretation (but not the statute itself) violated the non-delegation doctrine. Id., at 1038. We disagree.
In a delegation challenge, the constitutional question is whether the statute has delegated legislative power to the agency. Article I, § 1, of the Constitution vests “[a]ll legislative Powers herein granted ... in a Congress of the United States.” This text permits no delegation of those powers, Loving v. United States, 517 U. S. 748, 771 (1996); see id., at 776-777 (Scalia, J., concurring in part and concurring in judgment), and so we repeatedly have said that when Congress confers decisionmaking authority upon agencies Congress must “lay down by legislative act an intelligible principle to which the person or body authorized to [act] is directed to conform.” J. W. Hampton, Jr., & Co. v. United States, 276 U. S. 394, 409 (1928). We have never suggested that an agency can cure an unlawful delegation of legislative power by adopting in its discretion a limiting construction of the statute. Both Fahey v. Mallonee, 332 U. S. 245, 252-253 (1947), and Lichter v. United States, 334 U. S. 742, 783 (1948), mention agency regulations in the course of their nondelegation discussions, but Lichter did so because a subsequent Congress had incorporated the regulations into a revised version of the statute, ibid., and Fahey because the custom*473ary practices in the area, implicitly incorporated into the statute, were reflected in the regulations, 832 U. S., at 250. The idea that an agency can cure an unconstitutionally stand-ardless delegation of power by declining to exercise some of that power seems to us internally contradictory. The very choice of which portion of the power to exercise — that is to say, the prescription of the standard that Congress had omitted — would itself be an exercise of the forbidden legislative authority. Whether the statute delegates legislative power is a question for the courts, and an agency’s voluntary self-denial has no bearing upon the answer.
We agree with the Solicitor General that the text of § 109(b)(1) of the CAA at a minimum requires that “[flor a discrete set of pollutants and based on published air quality criteria that reflect the latest scientific knowledge, [the] EPA must establish uniform national standards at a level that is requisite to protect public health from the adverse effects of the pollutant in the ambient air.” Tr. of Oral Arg. ih-No. 99-1257, p. 5. Requisite, in turn, “mean[s] sufficient, but not more than necessary.” Id., at 7. These limits on the EPA’s discretion are strikingly similar to the ones we approved in Touby v. United States, 500 U. S. 160 (1991), which permitted the Attorney General to designate a drug as a controlled substance for purposes of criminal drug enforcement if doing so was “ ‘necessary to avoid an imminent hazard to the public safety.’” Id., at 163. They also resemble the Occupational Safety and Health Act of 1970 provision requiring the agency to “ ‘set the standard which most adequately assures, to the extent feasible, on the basis of the best available evidence, that no employee will suffer any impairment of health’ ” — which the Court upheld in Industrial Union Deyt., AFL-GIO v. American Petroleum Institute, 448 U. S. 607, 646 (1980), and which even then-JuSTlCE Rehnquist, who alone in that case thought the statute violated the nondelegation doctrine, see id., at 671 (opinion concurring in judgment), would have upheld if, like the statute *474here, it did not permit economic costs to be considered. See American Textile Mfrs. Institute, Inc. v. Donovan, 452 U. S. 490, 545 (1981) (Rehnquist, J., dissenting).
The scope of discretion § 109(b)(1) allows is in fact well within the outer limits of our nondelegation precedents. In the. history of the Court we have found the requisite “intelligible principle” lacking in only two statutes, one of which provided literally no guidance for the exercise of discretion, and the other of which conferred authority to regulate the entire economy on the basis of no more precise a standard than stimulating the economy by assuring “fair competition.” See Panama Refining Co. v. Ryan, 293 U. S. 388 (1935); A. L. A. Schechter Poultry Corp. v. United States, 295 U. S. 495 (1935). We have, on the other hand, upheld the validity of § 11(b)(2) of the Public Utility Holding Company Act of 1935, 49 Stat. 821, which gave the Securities and Exchange Commission authority to modify the structure of holding company systems so as to ensure that they are not “unduly or unnecessarily complicate^]” and do not “unfairly or inequitably distribute voting power among security holders.” American Power & Light Co. v. SEC, 329 U. S. 90, 104 (1946). We have approved the wartime conferral of agency power to fix the prices of commodities at a level that “ ‘will be generally fair and equitable and will effectuate the [in some respects conflicting] purposes of th[e] Act.’” Yakus v. United States, 321 U. S. 414, 420, 423-426 (1944). And we have found an “intelligible principle” in various statutes authorizing regulation in the “public interest,” See, e. g., National Broadcasting Co. v. United States, 319 U. S. 190, 225-226 (1943) (Federal Communications Commission’s power to regulate airwaves); New York Central Securities Corp. v. United States, 287 U. S. 12, 24-25 (1932) (Interstate Commerce Commission’s power to approve railroad consolidations). In short, we have “almost never felt qualified to second-guess Congress regarding the permissible degree of policy judgment that can be left to those executing or apply*475ing the law.” Mistretta v. United States, 488 U. S. 361, 416 (1989) (Scalia, J., dissenting); see id., at 373 (majority opinion).
It is true enough that the degree of agency discretion that is acceptable varies according to the scope of the power eon-gressionally conferred. See Loving v. United States, 517 U. S., at 772-773; United States v. Mazurie, 419 U. S. 544, 556-557 (1975). While Congress need not provide any direction to the EPA regarding the manner in which it is to define “country elevators,” which are to be exempt from new-stationary-source regulations governing grain elevators, see 42 U. S. C. §7411(i), it must provide substantial guidance on setting air standards that affect the entire national economy. But even in sweeping regulatory schemes we have never demanded, as the Court of Appeals did here, that statutes provide a “determinate criterion” for saying “how much [of the regulated harm] is too much.” 175 F. 3d, at 1034. In Touby, for example, we did not require the statute to decree how “imminent” was too imminent, or how “necessary” was necessary enough, or even — most relevant here — how “hazardous” was too hazardous. 500 U. S., at 165-167. Similarly, the statute at issue in Lichter authorized agencies to recoup “excess profits” paid under wartime Government contracts, yet we did not insist that Congress specify how much profit was too much. 334 U. S., at 783-786. It is therefore not conclusive for delegation purposes that, as respondents argue, ozone and particulate matter are “nonthreshold” pollutants that inflict a continuum of adverse health effects at any airborne concentration greater than zero, and hence require the EPA to make judgments of degree. “[A] certain degree of discretion, and thus of lawmaking, inheres in most executive or judicial action.” Mistretta v. United States, supra, at 417 (Scalia, J., dissenting) (emphasis deleted); see 488 U. S., at 378-379 (majority opinion). Section 109(b)(1) of the CAA, which to repeat we interpret as requiring the EPA to set air quality standards at the level that is “requi*476site” — that is, not lower or higher than is necessary — to protect the public health with an adequate margin of safety, fits comfortably within the scope of discretion permitted by our precedent.
We therefore reverse the judgment of the Court of Appeals remanding for reinterpretation that would avoid a supposed delegation of legislative power. It will remain for the Court of Appeals — on the remand that we direct for other reasons — to dispose of any other preserved challenge to the NAAQS under the judicial-review provisions contained in 42 U. S. C. § 7607(d)(9).
IV
The final two issues on which we granted certiorari concern the EPA’s authority to implement the revised ozone NAAQS in areas whose ozone levels currently exceed the maximum level permitted by that standard. The CAA designates such areas “nonattainment,” § 107(d)(1), 42 U. S. C. § 7407(d)(1); see also Pub. L. 105-178, §6103, 112 Stat. 465 (setting timeline for new ozone designations), and it exposes them to additional restrictions over and above the implementation requirements imposed generally by §110 of the CAA. These additional restrictions are found in the five substantive subparts of Part D of Title I, 42 U. S. C. §§7501-7515. Subpart 1, §§7501-7509a, contains general nonattainment regulations that pertain to every pollutant for which a NAAQS exists. Subparts 2 through 5, §§ 7511— 7514a, contain rules tailored to specific individual pollutants. Subpart 2, added by the Clean Air Act Amendments of 1990, §103, 104 Stat. 2423, addresses ozone. 42 U. S. C. §§ 7511— 75111 The dispute before us here, in a nutshell, is whether Subpart 1 alone (as the agency determined), or rather Sub-part 2 or some combination of Subparts 1 and 2, controls the implementation of the revised ozone NAAQS in non-attainment areas.
*477A
The Administrator first urges, however, that we vacate the judgment of the Court of Appeals on this issue because it lacked jurisdiction to review the EPA’s implementation policy. Section 307(b)(1) of the CAA, 42 U. S. C. § 7607(b)(1), gives the court jurisdiction over “any . . . nationally applicable regulations promulgated, or final action taken, by the Administrator,” but the EPA argues that its implementation policy was not agency “action,” was not “final” action, and is not ripe for review. We reject each of these three contentions.
At the same time the EPA proposed the revised ozone NAAQS in 1996, it also proposed an “interim implementation policy” for the NAAQS, see 61 Fed. Reg. 65752 (1996), that was to govern until the details of implementation could be put in final form through specific “rulemaking actions.” The preamble to this proposed policy declared that “the interim implementation policy . . . represent^] EPA’s preliminary views on these issues and, while it may include various statements that States must take certain actions, these statements are made pursuant to EPA’s preliminary interpretations, and thus do not bind the States and public as a matter of law.” Ibid. If the EPA had done no more, we perhaps could accept its current claim that its action was not final. However, after the agency had accepted comments on its proposed policy, and on the same day that the final ozone NAAQS was promulgated, the White House published in the Federal Register what it titled a “Memorandum for the Administrator of the Environmental Protection Agency” that prescribed implementation procedures for the EPA to follow. 62 Fed. Reg. 38421 (1997). (For purposes of our analysis we shall assume that this memorandum was not itself action by the EPA.) The EPA supplemented this memorandum with an explanation of the implementation procedures, which it published in the explanatory preamble to its final ozone *478NAAQS under the heading, “Final decision on the primary standard.” Id., at 38873. “In light of comments received regarding the interpretation proposed in the Interim Implementation Policy,” the EPA announced, it had “reconsidered that interpretation” and settled on a new one. Ibid. The provisions of “subpart 1 of part D of Title I of the Act” will immediately “apply to the implementation of the new 8-hour [ozone] standards.” Ibid.; see also id., at 38885 (new standard to be implemented “simultaneously [with the old standard] . . . under the provisions of . . . subpart 1”). Moreover, the provisions of subpart 2 “will [also] continue to apply as a matter of law for so long as an area is not attaining the [old] 1-hour standard.” Id., at 38873. Once the area reaches attainment for the old standard, however, “the provisions of subpart 2 will have been achieved and those provisions will no longer apply.” Ibid.; see also id., at 38884-38885.
We have little trouble concluding that this constitutes final agency action subject to review under §307. The bite in the phrase “final action” (which bears the same meaning in § 307(b)(1) that it does under the Administrative Procedure Act (APA), 5 U. S. C. § 704, see Harrison v. PPG Industries, Inc., 446 U. S. 578, 586 (1980)), is not in the word “action,” which is meant to cover comprehensively every manner in which an agency may exercise its power. See FTC v. Standard Oil Co. of Cal., 449 U. S. 232, 238, n. 7 (1980). It is rather in the word “final,” which requires that the action under review “mark the consummation of the agency’s de-cisionmaking process.” Bennett v. Spear, 520 U. S. 154, 177-178 (1997). Only if the “EPA has rendered its last word on the matter” in question, Harrison v. PPG Industries, Inc., supra, at 586, is its action “final” and thus reviewable. That standard is satisfied here. The EPA’s “decision-making process,” which began with the 1996 proposal and continued with the reception of public comments, concluded *479when the agency, “in light of [these comments],” and in conjunction with a corresponding directive from the White House, adopted the interpretation of Part D at issue here. Since that interpretation issued, the EPA has refused in subsequent rulemakings to reconsider it, explaining to disappointed commenters that its earlier decision was conclusive. See 63 Fed. Reg. 31014, 31018-31019 (1998). Though the agency has not dressed its decision with the conventional procedural accoutrements of finality, its own behavior thus belies the claim that its interpretation is not final.
The decision is also ripe for our review. “Ripeness ‘requires] us to evaluate both the fitness of the issues for judicial decision and the hardship to the parties of withholding court consideration.’ ” Texas v. United States, 523 U. S. 296, 300-301 (1998) (quoting Abbott Laboratories v. Gardner, 387 U. S. 136, 149 (1967)). The question before us here is purely one of statutory interpretation that would not “benefit from further factual development of the issues presented.” Ohio Forestry Assn., Inc. v. Sierra Club, 523 U. S. 726, 733 (1998). Nor will our review “inappropriately interfere with further administrative action,” ibid., since the EPA has concluded its consideration of the implementation issue. Finally, as for hardship to the parties: The respondent States must — on pain of forfeiting to the EPA control over implementation of the NAAQS — promptly undertake the lengthy and expensive task of developing state implementation plans (SIP’s) that will attain the new, more stringent standard within five years. See 42 U. S. C. §§7410, 7502. Whether or not this would suffice in an ordinary ease brought under the review provisions of the APA, see 5 U. S. C. § 704, we have characterized the special judicial-review provision of the CAA, 42 U. S. C. § 7607(b), as one of those statutes that specifically provides for “preenforcement” review, see Ohio Forestry Assn., Inc. v. Sierra Club, supra, at 737. Such statutes, we have said, permit “judicial review directly, even before the *480concrete effects normally required for APA review are felt.” Lujan v. National Wildlife Federation, 497 U. S. 871, 891 (1990). The effects at issue here surely meet that lower standard.
Beyond all this, the implementation issue was fairly included within the challenges to the final ozone rule that were properly before the Court of Appeals. Respondents argued below that the EPA could not revise the ozone standard, because to do so would trigger the use of Subpart 1, which had been supplanted (for ozone) by the specific rules of Sub-part 2. Brief for Industry Petitioners and Intervenors in No. 97-1441 (and consolidated cases) (CADC), pp. 82-84. The EPA responded that Subpart 2 did not supplant but simply supplemented Subpart 1, so that the latter section still “applies to all nonattainment areas for all NAAQS, . . . including nonattainment areas for any revised ozone standard.” Final Brief for EPA in No. 97-1441 (and consolidated cases) (CADC), pp. 67-68. The agency later reiterated that Subpart 2 “does not supplant implementation provisions for revised ozone standards. This interpretation fully harmonizes Subpart 2 with EPA’s clear authority to revise any NAAQS.” Id., at 71. In other words, the EPA was arguing that the revised standard could be issued, despite its apparent incompatibility with portions of S'ubpart 2, because it would be implemented under Subpart 1 rather than Subpart 2. The District of Columbia Circuit ultimately agreed that Subpart 2 could be harmonized with the EPA’s authority to promulgate revised NAAQS, but not because Subpart 2 is entirely inapplicable — which is one of EPA’s assignments of error. It is unreasonable to contend, as the EPA now does, that the Court of Appeals was obligated to reach the agency’s preferred result, but forbidden to assess the reasons the EPA had given for reaching that result. The implementation issue was fairly included within respondents’ challenge to the ozone rule, which all parties agree is final agency action ripe for review.
*481B
Our approach to the merits of the parties’ dispute is the familiar one of Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837 (1984). If the statute resolves the question whether Subpart 1 or Subpart 2 (or some combination of the two) shall apply to revised ozone NAAQS, then “that is the end of the matter.” Id., at 842-843. But if the statute is “silent or ambiguous” with respect to the issue, then we must defer to a “reasonable interpretation made by the administrator of an agency.” Id., at 844. We cannot agree with the Court of Appeals that Sub-part 2 clearly controls the implementation of revised ozone NAAQS, see 175 F. 3d, at 1048-1050, because we find the statute to some extent ambiguous. We conclude, however, that the agency’s interpretation goes beyond the limits of what is ambiguous and contradicts what in our view is quite clear. We therefore hold the implementation policy unlawful. See AT&T Corp. v. low a Utilities Bd., 525 U. S. 366, 392 (1999).
The text of Subpart 1 at first seems to point the way to a clear answer to the question, which Subpart controls? Two sections of Subpart 1, 7502(a)(1)(C) and 7502(a)(2)(D), contain switching provisions stating that if the classification of ozone nonattainment areas is “specifically provided [for] under other provisions of [Part D],” then those provisions will control instead of Subpart l’s. Thus, it is true but incomplete to note, as the Administrator does, that the substantive language of Subpart 1 is broad enough to apply to revised ozone standards. See, e. g., § 7502(a)(1)(A) (instructing the Administrator to classify nonattainment areas according to “any revised standard, including a revision of any standard in effect on November 15, 1990”); § 7502(a)(2)(A) (setting attainment deadlines). To determine whether that language does apply one must resolve the further textual issue whether some other provision, namely Subpart 2, provides for the classification of ozone nonattainment areas. If *482it does, then according to the switching provisions of Sub-part 1 it will control.
So, does Subpart 2 provide for classifying nonattainment ozone areas under the revised standard? It unquestionably does. The backbone of the subpart is Table 1, printed in § 7511(a)(1) and reproduced in the margin here,5 which defines five categories of ozone nonattainment areas and prescribes attainment deadlines for each. Section 7511(a)(1) funnels all nonattainment areas into the table for classification, declaring that “[e]ach area designated nonattainment for ozone . . . shall be classified at the time of such designation, under table 1, by operation of law.” And once an area has been classified, “the primary standard attainment date for ozone shall be as expeditiously as practicable but not later than the date provided in table 1.” The EPA argues that this text is not as clear or comprehensive as it seems, because the title of § 7511(a) reads “Classification and attainment dates for 1989 nonattainment areas,” which suggests that Subpart 2 applies only to areas that were in nonattainment in 1989, and not to areas later designated non-*483attainment under a revised ozone standard. The suggestion must be rejected, however, because § 7511(b)(1) specifically provides for the classification of areas that were in attainment in 1989 but have subsequently slipped into nonattainment. It thus makes clear that Subpart 2 is not limited solely to 1989 nonattainment areas. This eliminates the interpretive role of the title, which may only “she[d] light on some ambiguous word or phrase in the statute itself,” Carter v. United States, 530 U. S. 255, 267 (2000) (internal quotation marks omitted) (quoting Pennsylvania Dept. of Corrections v. Yeskey, 524 U. S., at 212, in turn quoting Trainmen v. Baltimore & Ohio R. Co., 331 U. S. 519, 528-529 (1947)).
It may well be, as the EPA argues — and as the concurring opinion below on denial of rehearing pointed out, see 195 F. 3d, at 11-12 — that some provisions of Subpart 2 are ill fitted to implementation of the revised standard. Using the old 1-hour averages of ozone levels, for example, as Sub-part 2 requires, see § 7511(a)(1); 44 Fed. Reg. 8202 (1979), would produce at best an inexact estimate of the new 8-hour averages, see 40 CFR §50.10, and App. I (1999). Also, to the extent that the new ozone standard is stricter than the old one, see Reply Brief for Petitioners in No. 99-1257, p. 17 (“the stricter 8-hour NAAQS”); 62 Fed. Reg. 38856, 38858 (1997) (8-hour standard of 0.09 ppm rather than 0.08 ppm would have “generally represented] the continuation of the [old] level of protection”), the classification system of Subpart 2 contains a gap, because it fails to classify areas whose ozone levels are greater than the new standard (and thus nonattaining) but less than the approximation of the old standard codified by Table 1. And finally, Subpart 2’s method for calculating attainment dates — which is simply to count forward a certain number of years from November 15, 1990 (the date the 1990 CAA Amendments took force), depending on how far out of attainment the area started — seems to make no sense for areas that are first classified under a new standard after November 15, 1990. *484If, for example, areas were classified in the year 2000, many of the deadlines would already have expired at the time of classification.
These gaps in Subpart 2's scheme prevent us from concluding that Congress clearly intended Subpart 2 to be the exclusive, permanent means of enforcing a revised ozone standard in nonattainment areas. The statute is in our view ambiguous concerning the manner in which Subpart 1 and Subpart 2 interact with regard to revised ozone standards, and we would defer to the EPA’s reasonable resolution of that ambiguity. See FDA v. Brown & Williamson Tobacco Corp., 529 U. S., at 132; INS v. Aguirre-Aguirre, 526 U. S. 415, 424 (1999). We cannot defer, however, to the interpretation the EPA has given.
Whatever effect may be accorded the gaps in Subpart 2 as implying some limited applicability of Subpart 1, they cannot be thought to render Subpart 2’s carefully designed restrictions on EPA discretion utterly nugatory once a new standard has been promulgated, as the EPA has concluded. The principal distinction between Subpart 1 and Subpart 2 is that the latter eliminates regulatory discretion that the former allowed. While Subpart 1 permits the EPA to establish classifications for nonattainment areas, Subpart 2 classifies areas as a matter of law based on a table. Compare § 7502(a)(1) with § 7511(a)(1) (Table 1). Whereas the EPA has discretion under Subpart 1 to extend attainment dates for as long as 12 years, under Subpart 2 it may grant no more than 2 years’ extension. Compare §§ 7502(a)(2)(A) and (C) with § 7511(a)(5). Whereas Subpart 1 gives the EPA considerable discretion to shape nonattainment programs, Subpart 2 prescribes large parts of them by law. Compare §§ 7502(c) and (d) with § 7511a. Yet according to the EPA, Subpart 2 was simply Congress’s “approach to the implementation of the [old] 1-hour” standard, and so there was no reason that “the new standard could not simultaneously be implemented under . . . subpart 1.” 62 Fed. Reg. *48538856, 38885 (1997); see also id., at 38873 (“[T]he provisions of subpart 1 . . . would apply to the implementation of the new 8-hour ozone standards”). To use a few apparent gaps in Subpart 2 to render its textually explicit applicability to nonattainment areas under the new standard utterly inoperative is to go over the edge of reasonable interpretation. The EPA may not construe the statute in a way that completely nullifies textually applicable provisions meant to limit its discretion.
The EPA’s interpretation making Subpart 2 abruptly obsolete is all the more astonishing because Subpart 2 was obviously written to govern implementation for some time. Some of the elements required to be included in SIP’s under Subpart 2 were not to take effect until many years after the passage of the CAA. See § 7511a(e)(3) (restrictions on “electric utility and industrial and commercial boilerfs]” to be “effective 8 years after November 15,1990”); §7511a(c)(5)(A) (vehicle monitoring program to “[b]egi[n] 6 years after November 15, 1990”); §7511a(g)(l) (emissions milestone requirements to be applied “6 years after November 15, 1990, and at intervals of every 3 years thereafter”). A plan reaching so far into the future was not enacted to be abandoned the next time the EPA reviewed the ozone standard — which Congress knew could happen at any time, since the technical staff papers had already been completed in late 1989. See 58 Fed. Reg. 13008, 13010 (1993); see also 42 U. S. C. § 7409(d)(1) (NAAQS must be reviewed and, if appropriate, revised at least once every five years). Yet nothing in the EPA’s interpretation would have prevented the agency from aborting Subpart 2 the day after it was enacted. Even now, if the EPA’s interpretation were correct, some areas of the country could be required to meet the new, more stringent ozone standard in at most the same time that Sub-part 2 had allowed them to meet the old standard. Compare § 7502(a)(2) (Subpart 1 attainment dates) with § 7511(a) (Sub-part 2 attainment dates). Los Angeles, for instance, “would *486be required to attain the revised NAAQS under Subpart 1 no later than the same year that marks the outer time limit for attaining Subpart 2’s one-hour ozone standard.” Brief for Petitioners in No. 99-1257, p. 49. An interpretation of Subpart 2 so at odds with its structure and manifest purpose cannot be sustained.
We therefore find the EPA’s implementation policy to be unlawful, though not in the precise respect determined by the Court of Appeals. After our remand, and the Court of Appeals’ final disposition of these cases, it is left to the EPA to develop a reasonable interpretation of the nonattainment implementation provisions insofar as they apply to revised ozone NAAQS.
* * *
To summarize our holdings in these unusually complex cases: (1) The EPA may not consider implementation costs in setting primary and secondary NAAQS under § 109(b) of the CAA. (2) Section 109(b)(1) does not delegate legislative power to the EPA in contravention of Art. I, § 1, of the Constitution. (3) The Court of Appeals had jurisdiction to review the EPA’s interpretation of Part D of Title I of the CAA, relating to the implementation of the revised ozone NAAQS. (4) The EPA’s interpretation of that Part is unreasonable.
The judgment of the Court of Appeals is affirmed in part and reversed in part, and the cases are remanded for proceedings consistent with this opinion.
It is so ordered.
None of the sections of the CAA in which the District of Columbia Circuit has found authority for the EPA to consider costs shares § 109(b)(l)’s prominence in the overall statutory scheme. See, e. g., Michigan v. EPA, 213 F. 3d 663, 678-679 (CADC 2000); George E. Warren Corp. v. EPA, 159 F. 3d 616, 623-624 (CADC 1998); Natural Resources Defense Council, Inc. v. EPA, 824 F. 2d 1146, 1154-1163 (CADC 1987) (en banc).
Respondents contend that this advice is required to be included in the NAAQS rulemaking record — which, if true, would suggest that it was relevant to the standard-setting process. But the provision respondents cite for their contention, 42 U. S. C. § 7607(d)(3), requires only that “pertinent findings, recommendations, and comments by the Scientific Review Committee” be included. The Committee’s advice concerning certain aspects of “adverse public health . .. effects” from various attainment strategies is unquestionably pertinent; but to say that Committee-generated cost data are pertinent is to beg the question. Likewise, while “all written comments” must be placed in the docket, § 7607(d)(4)(B)(i), the EPA need respond only to the “significant” ones, § 7407(d)(6)(B); comments regarding cost data are not significant if cost data are irrelevant.
Respondents scarcely mention in their arguments the secondary NAAQS required by § 109(b)(2), 42 U. S. C. § 7409(b)(2). For many of the same reasons described in the body of the opinion, as well as the text of § 109(b)(2), which instructs the EPA to set the standards at a level “requisite to protect the public welfare from any known or anticipated adverse effects associated with the presence of such air pollutant in the ambient air” (emphasis added), we conclude that the EPA may not consider implementation costs in setting the secondary NAAQS.
Respondents’ speculation that the EPA is secretly considering the costs of attainment without telling anyone is irrelevant to our interpretive inquiry. If such an allegation could be proved, it would be grounds for vacating the NAAQS, because the Administrator had not followed the law. See, e. g., Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842-843 (1984); Atlantic Mut. Ins. Co. v. Commissioner, 523 U. S. 382, 387 (1998). It would not, however, be grounds for this Court’s changing the law.
TABLE 1
Primary standard
Area class Design value* attainment date**
Marginal. 0.121 up to 0.138. 3 years after November 15,1990
Moderate. 0.138 up to 0.160. 6 years after November 15,1990
Serious. 0.160 up to 0.180. 9 years after November 15,1990
Severe. 0.180 up to 0.280. 15 years after November 15,1990
Extreme. 0.280 and above. 20 years after November 15,1990
*The design value is measured in parts per million (ppm).
**The primary standard attainment date is measured from November 15,1990.