Justice Stevens

delivered the opinion of the Court.

In the Clean Air Act Amendments of 1977, Pub. L. 95-95, 91 Stat. 685, Congress enacted certain requirements appli*840cable to States that had not achieved the national air quality standards established by the Environmental Protection Agency (EPA) pursuant to earlier legislation. The amended Clean Air Act required these “nonattainment” States to establish a permit program regulating “new or modified major stationary sources” of air pollution. Generally, a permit may not be issued for a new or modified major stationary source unless several stringent conditions are met.1 The EPA regulation promulgated to implement this permit requirement allows a State to adopt a plantwide definition of the term “stationary source.”2 Under this definition, an existing plant that contains several pollution-emitting devices may install or modify one piece of equipment without meeting the permit conditions if the alteration will not increase the total emissions from the plant. The question presented by these cases is whether EPA’s decision to allow States to treat all of the pollution-emitting devices within the same industrial grouping as though they were encased within a single “bubble” is based on a reasonable construction of the statutory term “stationary source.”

I

The EPA regulations containing the plantwide definition of the term stationary source were promulgated on October *84114, 1981. 46 Fed. Reg. 50766. Respondents3 filed a timely petition for review in the United States Court of Appeals for the District of Columbia Circuit pursuant to 42 U. S. C. 17607(b)(1).4 The Court of Appeals set aside the regulations. National Resources Defense Council, Inc. v. Gorsuch, 222 U. S. App. D. C. 268, 685 F. 2d 718 (1982).

The court observed that the relevant part of the amended Clean Air Act “does not explicitly define what Congress envisioned as a ‘stationary source, to which the permit program . . . should apply,” and further stated that the precise issue was not “squarely addressed in the legislative history.” Id., at 273, 685 F. 2d, at 723. In light of its conclusion that the legislative history bearing on the question was “at best contradictory,” it reasoned that “the purposes of the non-attainment program should guide our decision here.” Id., at 276, n. 39, 685 F. 2d, at 726, n. 39.5 Based on two of its precedents concerning the applicability of the bubble concept to certain Clean Air Act programs,6 the court stated that the bubble concept was “mandatory” in programs designed merely to maintain existing air quality, but held that it was “inappropriate” in programs enacted to improve air quality. Id., at 276, 685 F. 2d, at 726. Since the purpose of the per*842mit program — its “raison d’Ure,” in the court’s view — was to improve air quality, the court held that the bubble concept was inapplicable in these cases under its prior precedents. Ibid. It therefore set aside the regulations embodying the bubble concept as contrary to law. We granted certiorari to review that judgment, 461 U. S. 956 (1983), and we now reverse.

The basic legal error of the Court of Appeals was to adopt a static judicial definition of the term “stationary source” when it had decided that Congress itself had not commanded that definition. Respondents do not defend the legal reasoning of the Court of Appeals.7 Nevertheless, since this Court reviews judgments, not opinions,8 we must determine whether the Court of Appeals’ legal error resulted in an erroneous judgment on the validity of the regulations.

HH H-(

When a court reviews an agency s construction of the statute which it administers, it is confronted with 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, *843as well as the agency, must give effect to the unambiguously expressed intent of Congress.9 If, however, the court determines Congress has not directly addressed the precise question at issue, the court does not simply impose its own construction on the statute,10 as would be necessary in the absence of an administrative interpretation. Rather, 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.11

“The power of an administrative agency to administer a congressionally created . . . program necessarily requires the formulation of policy and the making of rules to fill any gap left, implicitly or explicitly, by Congress.” Morton v. Ruiz, 415 U. S. 199, 231 (1974). If Congress has explicitly left a gap for the agency to fill, there is an express delegation *844of authority to the agency to elucidate a specific provision of the statute by regulation. Such legislative regulations are given controlling weight unless they are arbitrary, capricious, or manifestly contrary to the statute.12 Sometimes the legislative delegation to an agency on a particular question is implicit rather than explicit. In such a case, a court may not substitute its own construction of a statutory provison for a reasonable interpretation made by the administrator of an agency.13

We have long recognized that considerable weight should be accorded to an executive department’s construction of a statutory scheme it is entrusted to administer,14 and the principle of deference to administrative interpretations

“has been consistently followed by this Court whenever decision as to the meaning or reach of a statute has involved reconciling conflicting policies, and a full understanding of the force of the statutory policy in the given situation has depended upon more than ordinary knowledge respecting the matters subjected to agency regulations. See, e. g., National Broadcasting Co. v. United States, 319 U. S. 190; Labor Board v. Hearst Publications, Inc., 322 U. S. 111; Republic Aviation Corp. v. *845Labor Board, 324 U. S. 793; Securities & Exchange Comm’n v. Chenery Corp., 332 U. S. 194; Labor Board v. Seven-Up Bottling Co., 344 U. S. 344.

. . If this choice represents a reasonable accommodation of conflicting policies that were committed to the agency’s care by the statute, we should not disturb it unless it appears from the statute or its legislative history that the accommodation is not one that Congress would have sanctioned.” United States v. Shimer, 367 U. S. 374, 382, 383 (1961).

Accord, Capital Cities Cable, Inc. v. Crisp, ante, at 699-700.

In light of these well-settled principles it is clear that the Court of Appeals misconceived the nature of its role in reviewing the regulations at issue. Once it determined, after its own examination of the legislation, that Congress did not actually have an intent regarding the applicability of the bubble concept to the permit program, the question before it was not whether in its view the concept is “inappropriate” in the general context of a program designed to improve air quality, but whether the Administrator’s view that it is appropriate in the context of this particular program is a reasonable one. Based on the examination of the legislation and its history which follows, we agree with the Court of Appeals that Congress did not have a specific intention on the applicability of the bubble concept in these cases, and conclude that the EPA’s use of that concept here is a reasonable policy choice for the agency to make.

Ill

In the 1950’s and the 1960’s Congress enacted a series of statutes designed to encourage and to assist the States in curtailing air pollution. See generally Train v. Natural Resources Defense Council, Inc., 421 U. S. 60, 63-64 (1975). The Clean Air Amendments of 1970, Pub. L. 91-604, 84 Stat. 1676, “sharply increased federal authority and responsibility *846in the continuing effort to combat air pollution,” 421 U. S., at 64, but continued to assign “primary responsibility for assuring air quality” to the several States, 84 Stat. 1678. Section 109 of the 1970 Amendments directed the EPA to promulgate National Ambient Air Quality Standards (NAAQS’s)15 and § 110 directed the States to develop plans (SIP’s) to implement the standards within specified deadlines. In addition, § 111 provided that major new sources of pollution would be required to conform to technology-based, performance standards; the EPA was directed to publish a list of categories of sources of pollution and to establish new source performance standards (NSPS) for each. Section 111(e) prohibited the operation of any new source in violation of a performance standard.

Section 111(a) defined the terms that are to be used in setting and enforcing standards of performance for new stationary sources. It provided:

“For purposes of this section:

“(3) The term ‘stationary source’ means any building, structure, facility, or installation which emits or may emit any air pollutant.” 84 Stat. 1683.

In the 1970 Amendments that definition was not only applicable to the NSPS program required by § 111, but also was made applicable to a requirement of § 110 that each state implementation plan contain a procedure for reviewing the location of any proposed new source and preventing its construction if it would preclude the attainment or maintenance of national air quality standards.16

In due course, the EPA promulgated NAAQS’s, approved SIP’s, and adopted detailed regulations governing NSPS’s *847for various categories of equipment. In one of its programs, the EPA used a plantwide definition of the term “stationary source.” In 1974, it issued NSPS’s for the nonferrous smelting industry that provided that the standards would not apply to the modification of major smelting units if their increased emissions were offset by reductions in other portions of the same plant.17

Nonattainment

The 1970 legislation provided for the attainment of primary NAAQS’s by 1975. In many areas of the country, particularly the most industrialized States, the statutory goals were not attained.18 In 1976, the 94th Congress was confronted with this fundamental problem, as well as many others respecting pollution control. As always in this area, the legislative struggle was basically between interests seeking strict schemes to reduce pollution rapidly to eliminate its social costs and interests advancing the economic concern that strict schemes would retard industrial development with attendant social costs. The 94th Congress, confronting these competing interests, was unable to agree on what response was in the public interest: legislative proposals to deal with nonattainment failed to command the necessary consensus.19

In light of this situation, the EPA published an Emissions Offset Interpretative Ruling in December 1976, see 41 Fed. Reg. 55524, to “fill the gap,” as respondents put it, until Congress acted. The Ruling stated that it was intended to *848address “the issue of whether and to what extent national air quality standards established under the Clean Air Act may restrict or prohibit growth of major new or expanded stationary air pollution sources.” Id., at 55524-55525. In general, the Ruling provided that “a major new source may locate in an area with air quality worse than a national standard only if stringent conditions can be met.” Id., at 55525. The Ruling gave primary emphasis to the rapid attainment of the statute’s environmental goals.20 Consistent with that emphasis, the construction of every new source in nonattainment areas had to meet the “lowest achievable emission rate” under the current state of the art for that type of facility. See Ibid. The 1976 Ruling did not, however, explicitly adopt or reject the “bubble concept.”21

IV

The Clean Air Act Amendments of 1977 are a lengthy, detailed, technical, complex, and comprehensive response to a major social issue. A small portion of the statute — 91 Stat. *849745-751 (Part D of Title I of the amended Act, 42 U. S. C. §§7501-7508) — expressly deals with nonattainment areas. The focal point of this controversy is one phrase in that portion of the Amendments.22

Basically, the statute required each State in a non-attainment area to prepare and obtain approval of a new SIP by July 1, 1979. In the interim those States were required to comply with the EPA’s interpretative Ruling of December 21, 1976. 91 Stat. 745. The deadline for attainment of the primary NAAQS’s was extended until December 31, 1982, and in some cases until December 31, 1987, but the SIP’s were required to contain a number of provisions designed to achieve the goals as expeditiously as possible.23

*850Most significantly for our purposes, the statute provided that each plan shall

“(6) require permits for the construction and operation of new or modified major stationary sources in accordance with section 173 . . . Id., at 747.

Before issuing a permit, § 173 requires (1) the state agency to determine that there will be sufficient emissions reductions in the region to offset the emissions from the new source and also to allow for reasonable further progress toward attainment, or that the increased emissions will not exceed an allowance for growth established pursuant to § 172(b)(5); (2) the applicant to certify that his other sources in the State are in compliance with the SIP, (3) the agency to determine that the applicable SIP is otherwise being implemented, and (4) the proposed source to comply with the lowest achievable emission rate (LAER).24

*851The 1977 Amendments contain no specific reference to the “bubble concept.” Nor do they contain a specific definition of the term “stationary source,” though they did not disturb the definition of “stationary source” contained in § 111(a)(3), applicable by the terms of the Act to the NSPS program. Section 302(j), however, defines the term “major stationary source” as follows:

“(j) Except as otherwise expressly provided, the terms ‘major stationary source’ and ‘major emitting facility’ mean any stationary facility or source of air pollutants which directly emits, or has the potential to emit, one hundred tons per year or more of any air pollutant (including any major emitting facility or source of fugitive emissions of any such pollutant, as determined by rule by the Administrator).” 91 Stat. 770.

V

The legislative history of the portion of the 1977 Amendments dealing with nonattainment areas does not contain any specific comment on the “bubble concept” or the question whether a plantwide definition of a stationary source is permissible under the permit program. It does, however, plainly disclose that in the permit program Congress sought to accommodate the conflict between the economic interest in permitting capital improvements to continue and the environmental interest in improving air quality. Indeed, the House Committee Report identified the economic interest as one of the “two main purposes” of this section of the bill. It stated:

“Section 117 of the bill, adopted during full committee markup establishes a new section 127 of the Clean Air Act. The section has two main purposes: (1) to allow reasonable economic growth to continue in an area while making reasonable further progress to assure attainment of the standards by a fixed date; and (2) to allow *852States greater flexibility for the former purpose than EPA’s present interpretative regulations afford.

“The new provision allows States with nonattainment areas to pursue one of two options. First, the State may proceed under EPA’s present ‘tradeoff’ or ‘offset’ ruling. The Administrator is authorized, moreover, to modify or amend that ruling in accordance with the intent and purposes of this section.

“The State’s second option would be to revise its implementation plan in accordance with this new provision.” H. R. Rep. No. 95-294, p. 211 (1977).25

The portion of the Senate Committee Report dealing with nonattainment areas states generally that it was intended to “supersede the EPA administrative approach,” and that expansion should be permitted if a State could “demonstrate that these facilities can be accommodated within its overall plan to provide for attainment of air quality standards.” S. Rep. No. 95-127, p. 55 (1977). The Senate Report notes the value of “case-by-case review of each new or modified major source of pollution that seeks to locate in a region exceeding an ambient standard,” explaining that such a review “requires matching reductions from existing sources against *853emissions expected from the new source in order to assure that introduction of the new source will not prevent attainment of the applicable standard by the statutory deadline.” Ibid. This description of a case-by-case approach to plant additions, which emphasizes the net consequences of the construction or modification of a new source, as well as its impact on the overall achievement of the national standards, was not, however, addressed to the precise issue raised by these cases.

Senator Mustíe made the following remarks:

“I should note that the test for determining whether a new or modified source is subject to the EPA interpretative regulation [the Offset Ruling] — and to the permit requirements of the revised implementation plans under the conference bill — is whether the source will emit a pollutant into an area which is exceeding a national ambient air quality standard for that pollutant — or precursor. Thus, a new source is still subject to such requirements as ‘lowest achievable emission rate’ even if it is constructed as a replacement for an older facility resulting in a net reduction from previous emission levels.

“A source — including an existing facility ordered to convert to coal — is subject to all the nonattainment requirements as a modified source if it makes any physical change which increases the amount of any air pollutant for which the standards in the area are exceeded.” 123 Cong. Rec. 26847 (1977).

I — I >

As previously noted, prior to the 1977 Amendments, the EPA had adhered to a plantwide definition of the term “source” under a NSPS program. After adoption of the 1977 Amendments, proposals for a plantwide definition were considered in at least three formal proceedings.

In January 1979, the EPA considered the question whether the same restriction on new construction in nonattainment areas that had been included in its December 1976 Ruling *854should be required in the revised SIP’s that were scheduled to go into effect in July 1979. After noting that the 1976 Ruling was ambiguous on the question “whether a plant with a number of different processes and emission.points would be considered a single source,” 44 Fed. Reg. 3276 (1979), the EPA, in effect, provided a bifurcated answer to that question. In those areas that did not have a revised SIP in effect by July 1979, the EPA rejected the plantwide definition; on the other hand, it expressly concluded that the plantwide approach would be permissible in certain circumstances if authorized by an approved SIP. It stated:

“Where a state implementation plan is revised and implemented to satisfy the requirements of Part D, including the reasonable further progress requirement, the plan requirements for major modifications may exempt modifications of existing facilities that are accompanied by intrasource offsets so that there is no net increase in emissions. The agency endorses such exemptions, which would provide greater flexibility to sources to effectively manage their air emissions at least cost.” Ibid.26

*855In April, and again in September 1979', the EPA published additional comments in which it indicated that revised SIP’s could adopt the plantwide definition of source in non-attainment areas in certain circumstances. See id., at 20372, 20379, 51924, 51951, 51958. On the latter occasion, the EPA made a formal rulemaking proposal that would have permitted the use of the “bubble concept” for new installations within a plant as well as for modifications of existing units. It explained:

“‘Bubble’ Exemption: The use of offsets inside the same source is called the ‘bubble.’ EPA proposes use of the definition of ‘source’ (see above) to limit the use of the bubble under nonattainment requirements in the following respects:

“i. Part D SIPs that include all requirements needed to assure reasonable further progress and attainment by the deadline under section 172 and that are being carried out need not restrict the use of a plantwide bubble, the same as under the PSD proposal.

“ii. Part D SIPs that do not meet the requirements specified must limit use of the bubble by including a definition of ‘installation’ as an identifiable piece of process equipment.”27

*856Significantly, the EPA expressly noted that the word “source” might be given a plantwide definition for some purposes and a narrower definition for other purposes. It wrote:

“Source means any building structure, facility, or installation which emits or may emit any regulated pollutant. ‘Building, structure, facility or installation’ means plant in PSD areas and in nonattainment areas except where the growth prohibitions would apply or where no adequate SIP exists or is being carried out.” Id., at 51925.28

The EPA’s summary of its proposed Ruling discloses a flexible rather than rigid definition of the term “source” to implement various policies and programs:

“In summary, EPA is proposing two different ways to define source for different kinds of NSR programs:

“(1) For PSD and complete Part D SIPs, review would apply only to plants, with an unrestricted plant-wide bubble.

“(2) For the offset ruling, restrictions on construction, and incomplete Part D SIPs, review would apply to both plants and individual pieces of process equipment, causing the plant-wide bubble not to apply for new and modified major pieces of equipment.

“In addition, for the restrictions on construction, EPA is proposing to define ‘major modification’ so as to prohibit the bubble entirely. Finally, an alternative discussed but not favored is to have only pieces of process equipment reviewed, resulting in no plant-wide bubble and allowing minor pieces of equipment to escape NSR *857regardless of whether they are within a major plant.” Id., at 51934.

In August 1980, however, the EPA adopted a regulation that, in essence, applied the basic reasoning of the Court of Appeals in these cases. The EPA took particular note of the two then-recent Court of Appeals decisions, which had created the bright-line rule that the “bubble concept” should be employed in a program designed to maintain air quality but not in one designed to enhance air quality. Relying heavily on those cases,29 EPA adopted a dual definition of “source” for nonattainment areas that required a permit whenever a change in either the entire plant, or one of its components, would result in a significant increase in emissions even if the increase was completely offset by reductions elsewhere in the plant. The EPA expressed the opinion that this interpretation was “more consistent with congressional intent” than the plantwide definition because it “would bring in more sources or modifications for review,” 45 Fed. Reg. 52697 (1980), but its primary legal analysis was predicated on the two Court of Appeals decisions.

In 1981 a new administration took office and initiated a “Government-wide reexamination of regulatory burdens and complexities.” 46 Fed. Reg. 16281. In the context of that *858review, the EPA reevaluated the various arguments that had been advanced in connection with the proper definition of the term “source” and concluded that the term should be given the same definition in both nonattainment areas and PSD areas.

In explaining its conclusion, the EPA first noted that the definitional issue was not squarely addressed in either the statute or its legislative history and therefore that the issue involved an agency “judgment as how to best carry out the Act.” Ibid. It then set forth several reasons for concluding that the plantwide definition was more appropriate. It pointed out that the dual definition “can act as a disincentive to new investment and modernization by discouraging modifications to existing facilities” and “can actually retard progress in air pollution control by discouraging replacement of older, dirtier processes or pieces of equipment with new, cleaner ones.” Ibid. Moreover, the new definition “would simplify EPA’s rules by using the same definition of ‘source’ for PSD, nonattainment new source review and the construction moratorium. This reduces confusion and inconsistency. ” Ibid. Finally, the agency explained that additional requirements that remained in place would accomplish the fundamental purposes of achieving attainment with NAAQS’s as expeditiously as possible.30 These conclusions were ex*859pressed in a proposed rulemaking in August 1981 that was formally promulgated in October. See id., at 50766.

VII

In this Court respondents expressly reject the basic rationale of the Court of Appeals’ decision. That court viewed the statutory definition of the term “source” as sufficiently flexible to cover either a plantwide definition, a narrower definition covering each unit within a plant, or a dual definition that could apply to both the entire “bubble” and its components. It interpreted the policies of the statute, however, to mandate the plantwide definition in programs designed to maintain clean air and to forbid it in programs designed to improve air quality. Respondents place a fundamentally different construction on the statute. They contend that the text of thé Act requires the EPA to use a dual definition — if either a component of a plant, or the plant as a whole, emits over 100 tons of pollutant, it is a major stationary source. They thus contend that the EPA rules adopted in 1980, insofar as they apply to the maintenance of the quality of clean air, as well as the 1981 rules which apply to nonattainment areas, violate the statute.31

Statutory Language

The definition of the term “stationary source” in § 111(a)(3) refers to “any building, structure, facility, or installation” which emits air pollution. See supra, at 846. This definition is applicable only to the NSPS program by the express terms of the statute; the text of the statute does not make this defi*860nition applicable to the permit program. Petitioners therefore maintain that there is no statutory language even relevant to ascertaining the meaning of stationary source in the permit program aside from §302(j), which defines the term “major stationary source.” See supra, at 851. We disagree with petitioners on this point.

The definition in §302(j) tells us what the word “major” means — a source must emit at least 100 tons of pollution to qualify — but it sheds virtually no light on the meaning of the term “stationary source.” It does equate a source with a facility — a “major emitting facility” and a “major stationary source” are synonymous under § 302(j). The ordinary meaning of the term “facility” is some collection of integrated elements which has been designed and constructed to achieve some purpose. Moreover, it is certainly no affront to common English usage to take a reference to a major facility or a major source to connote an entire plant as opposed to its constituent parts. Basically, however, the language of §302(j) simply does not compel any given interpretation of the term “source.”

Respondents recognize that, and hence point to § 111(a)(3). Although the definition in that section is not literally applicable to the permit program, it sheds as much light on the meaning of the word “source” as anything in the statute.32 As respondents point out, use of the words “building, structure, facility, or installation,” as the definition of source, could be read to impose the permit conditions on an individual building that is a part of a plant.33 A “word may have a character of its own not to be submerged by its association.” Russell Motor Car Co. v. United States, 261 U. S. 514, 519 *861(1923). On the other hand, the meaning of a word must be ascertained in the context of achieving particular objectives, and the words associated with it may indicate that the true meaning of the series is to convey a common idea. The language may reasonably be interpreted to impose the requirement on any discrete, but integrated, operation which pollutes. This gives meaning to all of the terms — a single building, not part of a larger operation, would be covered if it emits more than 100 tons of pollution, as would any facility, structure, or installation. Indeed, the language itself implies a “bubble concept” of sorts: each enumerated item would seem to be treated as if it were encased in a bubble. While respondents insist that each of these terms must be given a discrete meaning, they also argue that § 111(a)(3) defines “source” as that term is used in § 302(j). The latter section, however, equates a source with a facility, whereas the former defines “source” as a facility, among other items.

We are not persuaded that parsing of general terms in the text of the statute will reveal an actual intent of Congress.34 *862We know full well that this language is not dispositive; the terms are overlapping and the language is not precisely-directed to the question of the applicability of a given term in the context of a larger operation. To the extent any congressional “intent” can be discerned from this language, it would appear that the listing of overlapping, illustrative terms was intended to enlarge, rather than to confine, the scope of the agency’s power to regulate particular sources in order to effectuate the policies of the Act.

Legislative History

In addition, respondents argue that the legislative history and policies of the Act foreclose the plantwide definition, and that the EPA’s interpretation is not entitled to deference because it represents a sharp break with prior interpretations of the Act.

Based on our examination of the legislative history, we agree with the Court of Appeals that it is unilluminating. The general remarks pointed to by respondents “were obviously not made with this narrow issue in mind and they cannot be said to demonstrate a Congressional desire . . . .” Jewell Ridge Coal Corp. v. Mine Workers, 325 U. S. 161, 168-169 (1945). Respondents’ argument based on the legislative history relies heavily on Senator Muskie’s observation that a new source is subject to the LAER requirement.35 But the full statement is ambiguous and like the text of § 173 itself, this comment does not tell us what a new source is, much less that it is to have an inflexible definition. We find that the legislative history as a whole is silent on the precise issue before us. It is, however, consistent with the view that the EPA should have broad discretion in implementing the policies of the 1977 Amendments.

*863More importantly, that history plainly identifies the policy concerns that motivated the enactment; the plantwide definition is fully consistent with one of those concerns — the allowance of reasonable economic growth — and, whether or not we believe it most effectively implements the other, we must recognize that the EPA has advanced a reasonable explanation for its conclusion that the regulations serve the environmental objectives as well. See supra, at 857-859, and n. 29; see also supra, at 855, n. 27. Indeed, its reasoning is supported by the public record developed in the rulemaking process,36 as well as by certain private studies.37

Our review of the EPA’s varying interpretations of the word “source” — both before and after the 1977 Amendments — convinces us that the agency primarily responsible for administering this important legislation has consistently interpreted it flexibly — not in a sterile textual vacuum, but in the context of implementing policy decisions in a technical and complex arena. The fact that the agency has from time to time changed its interpretation of the term “source” does not, as respondents argue, lead us to conclude that no deference should be accorded the agency’s interpretation of the statute. An initial agency interpretation is not instantly carved in stone. On the contrary, the agency, to engage in informed rulemaking, must consider varying interpretations *864and the wisdom of its policy on a continuing basis. Moreover, the fact that the agency has adopted different definitions in different contexts adds force to the argument that the definition itself is flexible, particularly since Congress has never indicated any disapproval of a flexible reading of the statute.

Significantly, it was not the agency in 1980, but rather the Court of Appeals that read the statute inflexibly to command a plantwide definition for programs designed to maintain clean air and to forbid such a definition for programs designed to improve air quality. The distinction the court drew may well be a sensible one, but our labored review of the problem has surely disclosed that it is not a distinction that Congress ever articulated itself, or one that the EPA found in the statute before the courts began to review the legislative work product. We conclude that it was the Court of Appeals, rather than Congress or any of the decision-makers who are authorized by Congress to administer this legislation, that was primarily responsible for the 1980 position taken by the agency.

Policy

The arguments over policy that are advanced in the parties’ briefs create the impression that respondents are now waging in a judicial forum a specific policy battle which they ultimately lost in the agency and in the 32 jurisdictions opting for the “bubble concept,” but one which was never waged in the Congress. Such policy arguments are more properly addressed to legislators or administrators, not to judges.38

*865In these cases the Administrator’s interpretation represents a reasonable accommodation of manifestly competing interests and is entitled to deference: the regulatory scheme is technical and complex,39 the agency considered the matter in a detailed and reasoned fashion,40 and the decision involves reconciling conflicting policies.41 Congress intended to accommodate both interests, but did not do so itself on the level of specificity presented by these cases. Perhaps that body consciously desired the Administrator to strike the balance at this level, thinking that those with great expertise and charged with responsibility for administering the provision would be in a better position to do so; perhaps it simply did not consider the question at this level; and perhaps Congress was unable to forge a coalition on either side of the question, and those on each side decided to take their chances with the scheme devised by the agency. For judicial purposes, it matters not which of these things occurred.

Judges are not experts in the field, and are not part of either political branch of the Government. Courts must, in some cases, reconcile competing political interests, but not on the basis of the judges’ personal policy preferences. In contrast, an agency to which Congress has delegated policymak-ing responsibilities may, within the limits of that delegation, properly rely upon the incumbent administration’s views of wise policy to inform its judgments. While agencies are not directly accountable to the people, the Chief Executive is, and it is entirely appropriate for this political branch of the Government to make such policy choices — resolving the competing interests which Congress itself either inadvertently did not resolve, or intentionally left to be resolved by the *866agency charged with the administration of the statute in light of everyday realities.

When a challenge to an agency construction of a statutory provision, fairly conceptualized, really centers on the wisdom of the agency’s policy, rather than whether it is a reasonable choice within a gap left open by Congress, the challenge must fail. In such a case, federal judges — who have no constituency — have a duty to respect legitimate policy choices made by those who do. The responsibilities for assessing the wisdom of such policy choices and resolving the struggle between competing views of the public interest are not judicial ones: “Our Constitution vests such responsibilities in the political branches.” TVA v. Hill, 437 U. S. 153, 195 (1978).

We hold that the EPA’s definition of the term “source” is a permissible construction of the statute which seeks to accommodate progress in reducing air pollution with economic growth. “The Regulations which the Adminstrator has adopted provide what the agency could allowably view as . .. [an] effective reconciliation of these twofold ends .. . .” United States v. Shimer, 367 U. S., at 383.

The judgment of the Court of Appeals is reversed.

It is so ordered.

Justice Marshall and Justice Rehnquist took no part in the consideration or decision of these cases.

Justice O’Connor took no part in the decision of these cases.