delivered the opinion of the Court.
These cases concern a set of regulations adopted by the Environmental Protection Agency (EPA or agency) under § 316(b) of the Clean Water Act, 33 U.S.C. § 1326(b). 69 Fed. Reg. 41576 (2004). Respondents — environmental groups and various States1 — challenged those regulations, and the Second Circuit set them aside. Riverkeeper, Inc. v. EPA, 475 F. 3d 83, 99-100 (2007). The issue for our decision is whether, as the Second Circuit held, the EPA is not permitted to use cost-benefit analysis in determining the content of regulations promulgated under § 1326(b).
I
Petitioners operate — or represent those who operate— large powerplants. In the course of generating power, those *213plants also generate large amounts of heat. To cool their facilities, petitioners employ “cooling water intake structures” that extract water from nearby water sources. These structures pose various threats to the environment, chief among them the squashing against intake screens (elegantly called “impingement”) or suction into the cooling system (“entrainment”) of aquatic organisms that live in the affected water sources. See 69 Fed. Reg. 41586. Accordingly, the facilities are subject to regulation under the Clean Water Act, 33 U. S. C. § 1251 et seq., which mandates:
“Any standard established pursuant to section 1311 of this title or section 1316 of this title and applicable to a point source shall require that the location, design, construction, and capacity of cooling water intake structures reflect the best technology available for minimizing adverse environmental impact.” § 1326(b).
Sections 1311 and 1316, in turn, employ a variety of “best technology” standards to regulate the discharge of effluents into the Nation’s waters.
The § 1326(b) regulations at issue here were promulgated by the EPA after nearly three decades in which the determination of the “best technology available for minimizing [cooling water intake structures’] adverse environmental impact” was made by permit-issuing authorities on a case-by-case basis, without benefit of a governing regulation. The EPA’s initial attempt at such a regulation came to nought when the Fourth Circuit determined that the agency had failed to adhere to the procedural requirements of the Administrative Procedure Act. Appalachian Power Co. v. Train, 566 F. 2d 451, 457 (1977). The EPA withdrew the regulation, 44 Fed. Reg. 32956 (1979), and instead published “draft guidance” for use in implementing § 1326(b)’s requirements via site-specific permit decisions under §1342. See EPA, Office of Water Enforcement Permits Div., {Draft} Guidance for Evaluating the Adverse Impact of Cooling *214Water Intake Structures on the Aquatic Environment: Section 316(b) P. L. 92-500 (May 1, 1977), online at http:// www.epa.gov/waterscience/316b/files/1977AEIguid.pdf (all Internet materials as visited Mar. 30, 2009, and available in Clerk of Court’s case file); 69 Fed. Reg. 41584 (describing system of case-by-case permits under the draft guidance).
In 1995, the EPA entered into a consent decree which, as subsequently amended, set a multiphase timetable for the EPA to promulgate regulations under § 1326(b). See Riverkeeper, Inc. v. Whitman, No. 93 Civ. 0314 (AGS), 2001 WL 1505497, *1 (SDNY, Nov. 27, 2001). In the first phase the EPA adopted regulations governing certain new, large cooling water intake structures. 66 Fed. Reg. 65256 (2001) (Phase I rules); see 40 CFR §§ 125.80(a), 125.81(a) (2008). Those rules require new facilities with water-intake flow greater than 10 million gallons per day to, among other things, restrict their inflow “to a level commensurate with that which can be attained by a closed-cycle recirculating cooling water system.”2 § 125.84(b)(1). New facilities with water-intake flow between 2 million and 10 million gallons per day may alternatively comply by, among other things, reducing the volume and velocity of water removal to certain levels. § 125.84(c). And all facilities may alternatively comply by demonstrating, among other things, “that the technologies employed will reduce the level of adverse environmental impact... to a comparable level” to what would be achieved by using a closed-cycle cooling system. § 125.84(d). These regulations were upheld in large part by the Second Circuit in Riverkeeper, Inc. v. EPA, 358 F. 3d 174 (2004).
*215The EPA then adopted the so-called “Phase II” rules at issue here.3 69 Fed. Reg. 41576. They apply to existing facilities that are point sources, whose primary activity is the generation and transmission (or sale for transmission) of electricity, and whose water-intake flow is more than 50 million gallons of water per day, at least 25 percent of which is used for cooling purposes. Ibid. Over 500 facilities, accounting for approximately 53 percent of the Nation’s electric-power generating capacity, fall within Phase II’s ambit. See EPA, Economic and Benefits Analysis for the Final Section 316(b) Phase II Existing Facilities Rule, p. A3-13 (Table A3-4, Feb. 2004), online at http://www. epa.gov/waterscience/316b/phase2/econbenefits/final/a3. pdf. Those facilities remove on average more than 214 billion gallons of water per day, causing impingement and entrainment of over 3.4 billion aquatic organisms per year. 69 Fed. Reg. 41586.
To address those environmental impacts, the EPA set “national performance standards,” requiring Phase II facilities (with some exceptions) to reduce “impingement mortality for all life stages of fish and shellfish by 80 to 95 percent from the calculation baseline”; a subset of facilities must also reduce entrainment of such aquatic organisms by “60 to 90 percent from the calculation baseline.” 40 CFR § 125.94(b)(1), (2); see § 125.93 (defining “calculation baseline”). Those targets are based on the environmental improvements achievable through deployment of a mix of remedial technologies, 69 Fed. Reg. 41599, which the EPA determined were “commercially available and economically practicable,” id., at 41602.
In its Phase II rules, however, the EPA expressly declined to mandate adoption of closed-cycle cooling systems or equiv*216alent reductions in impingement and entrainment, as it had done for new facilities subject to the Phase I rules. Id., at 41601. It refused to take that step in part because of the “generally high costs” of converting existing facilities to closed-cycle operation, and because “other technologies approach the performance of this option.” Id., at 41605. Thus, while closed-cycle cooling systems could reduce impingement and entrainment mortality by up to 98 percent, id., at 41601 (compared to the Phase II targets of 80 to 95 percent impingement reduction), the cost of rendering all Phase II facilities closed-cycle-compliant would be approximately $3.5 billion per year, id., at 41605, nine times the estimated cost of compliance with the Phase II performance standards, id., at 41666. Moreover, Phase II facilities compelled to convert to closed-cycle cooling systems “would produce 2.4 percent to 4.0 percent less electricity even while burning the same amount of coal,” possibly requiring the construction of “20 additional 400-MW plants ... to replace the generating capacity lost.” Id., at 41605. The EPA thus concluded that “[although not identical, the ranges of impingement and entrainment reduction are similar under both options .... [Benefits of compliance with the Phase II rules] can approach those of closed-cycle recirculating systems at less cost with fewer implementation problems.” Id., at 41606.
The regulations permit the issuance of site-specific variances from the national performance standards if a facility can demonstrate either that the costs of compliance are “significantly greater than” the costs considered by the agency in setting the standards, 40 CFR § 125.94(a)(5)(i), or that the costs of compliance “would be significantly greater than the benefits of complying with the applicable performance standards,” § 125.94(a)(5)(h). Where a variance is warranted, the permit-issuing authority must impose remedial measures that yield results “as close as practicable to the applicable performance standards.” § 125.94(a)(5)(i), (ii).
*217Respondents challenged the EPA’s Phase II regulations, and the Second Circuit granted their petition for review and remanded the regulations to the EPA. The Second Circuit identified two ways in which the EPA could permissibly consider costs under 33 U. S. C. § 1326(b): (1) in determining whether the costs of remediation “can be ‘reasonably borne’ by the industry,” and (2) in determining which remedial technologies are the most cost effective, that is, the technologies that reach a specified level of benefit at the lowest cost. 475 F. 3d, at 99-100. See also id., at 98, and n. 10. It concluded, however, that cost-benefit analysis, which “compares the costs and benefits of various ends, and chooses the end with the best net benefits,” id., at 98, is impermissible under § 1326(b), id., at 100.
The Court of Appeals held the site-specific cost-benefit variance provision to be unlawful. Id., at 114. Finding it unclear whether the EPA had relied on cost-benefit analysis in setting the national performance standards, or had only used cost-effectiveness analysis, it remanded to the agency for clarification of that point. Id., at 104-105. (The remand was also based on other grounds which are not at issue here.) The EPA suspended operation of the Phase II rules pending further rulemaking. 72 Fed. Reg. 37107 (2007). We then granted certiorari limited to the following question: “Whether [§ 1326(b)] . . . authorizes the [EPA] to compare costs with benefits in determining ‘the best technology available for minimizing adverse environmental impact’ at cooling water intake structures.” 552 U. S. 1309 (2008).
II
In setting the Phase II national performance standards and providing for site-specific cost-benefit variances, the EPA relied on its view that §1326(b)’s “best technology available” standard permits consideration of the technology’s costs, 69 Fed. Reg. 41626, and of the relationship between those costs and the environmental benefits produced, id., at *21841603. That view governs if it is a reasonable interpretation of the statute — not necessarily the only possible interpretation, nor even the interpretation deemed most reasonable by the courts. Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 843-844 (1984).4
As we have described, § 1326(b) instructs the EPA to set standards for cooling water intake structures that reflect “the best technology available for minimizing adverse environmental impact.” The Second Circuit took that language to mean the technology that achieves the greatest reduction in adverse environmental impacts at a cost that can reasonably be borne by the industry. 475 F. 3d, at 99-100. That is certainly a plausible interpretation of the statute. The “best” technology — that which is “most advantageous,” Webster’s New International Dictionary 258 (2d ed. 1953) — may well be the one that produces the most of some good, here a reduction in adverse environmental impact. But “best technology” may also describe the technology that most efficiently produces some good. In common parlance one could certainly use the phrase “best technology” to refer to that which produces a good at the lowest per-unit cost, even if it produces a lesser quantity of that good than other available technologies.
Respondents contend that this latter reading is precluded by the statute’s use of the phrase “for minimizing adverse *219environmental impact.” Minimizing, they argue, means reducing to the smallest amount possible, and the “best technology available for minimizing adverse environmental impacts” must be the economically feasible technology that achieves the greatest possible reduction in environmental harm. Brief for Respondent Riverkeeper, Inc., et al. 25-26. But “minimize” is a term that admits of degree and is not necessarily used to refer exclusively to the “greatest possible reduction.” For example, elsewhere in the Clean Water Act, Congress declared that the procedures implementing the Act “shall encourage the drastic minimization of paperwork and interagency decision procedures.” 33 U. S. C. § 1251(f). If respondents’ definition of the term “minimize” is correct, the statute’s use of the modifier “drastic” is superfluous.
Other provisions in the Clean Water Act also suggest the agency’s interpretation. When Congress wished to mandate the greatest feasible reduction in water pollution, it did so in plain language: The provision governing the discharge of toxic pollutants into the Nation’s waters requires the EPA to set “effluent limitations [which] shall require the elimination of discharges of all pollutants if the Administrator finds ... that such elimination is technologically and economically achievable,” § 1311(b)(2)(A) (emphasis added). See also § 1316(a)(1) (mandating “where practicable, a standard [for new point sources] permitting no discharge of pollutants” (emphasis added)). Section 1326(b)’s use of the less ambitious goal of “minimizing adverse environmental impact” suggests, we think, that the agency retains some discretion to determine the extent of reduction that is warranted under the circumstances. That determination could plausibly involve a consideration of the benefits derived from reductions and the costs of achieving them. Cf. 40 CFR § 125.83 (defining “minimize” for purposes of the Phase I regulations as “reducing] to the smallest amount, extent, or degree reasonably possible”). It seems to us, therefore, that the phrase *220“best technology available,” even with the added specification “for minimizing adverse environmental impact,” does not unambiguously preclude cost-benefit analysis.5
Respondents’ alternative (and, alas, also more complex) argument rests upon the structure of the Clean Water Act. The Act provided that during its initial implementation period existing “point sources” — discrete conveyances from which pollutants are or may be discharged, 33 U. S. C. § 1362(14) — were subject to “effluent limitations . . . which shall require the application of the best practicable control technology currently available.” § 1311(b)(1)(A) (emphasis added). (We shall call this the “BPT” test.) Following that transition period, the Act initially mandated adoption, by July 1, 1983 (later extended to March 31, 1989), of stricter effluent limitations requiring “application of the best available technology economically achievable for such category or class, which will result in reasonable further progress toward the national goal of eliminating the discharge of all pollutants.” § 1311(b)(2)(A) (emphasis added); see EPA v. National Crushed Stone Assn., 449 U. S. 64, 69-70 (1980). (We shall call this the “BATEA” test.) Subsequent amendment limited application of this standard to toxic and noneonventional pollutants, and for the remainder established a (presumably laxer) test of “best conventional-pollutant control technology.” § 1311(b)(2)(E).6 (We shall call this “BCT.”) *221Finally, §1316 subjected certain categories of new point sources to “the greatest degree of effluent reduction which the Administrator determines to be achievable through application of the best available demonstrated control technology. ” § 1316(a)(1) (emphasis added); § 1316(b)(1)(B). (We shall call this the “BADT” test.) The provision at issue here, applicable not to effluents but to cooling water intake structures, requires, as we have described, “the best technology available for minimizing adverse environmental impact,” § 1326(b) (emphasis added). (We shall call this the “BTA” test.)
The first four of these tests are elucidated by statutory factor lists that guide their implementation. To take the standards in (presumed) order of increasing stringency, see Crushed Stone, supra, at 69-70: In applying the BPT test the EPA is instructed to consider, among other factors, “the total cost of application of technology in relation to the effluent reduction benefits to be achieved.” § 1314(b)(1)(B). In applying the BCT test it is instructed to consider “the reasonableness of the relationship between the costs of attaining a reduction in effluents and the effluent reduction benefits derived.” § 1314(b)(4)(B) (emphasis added). And in applying the BATEA and BADT tests the EPA is instructed to consider the “cost of achieving such effluent reduction.” §§ 1314(b)(2)(B), 1316(b)(1)(B). There is no such elucidating language applicable to the BTA test at issue here. To facilitate comparison, the texts of these five tests, the clarifying factors applicable to them, and the entities to which they apply are set forth in the Appendix, infra.
The Second Circuit, in rejecting the EPA’s use of cost-benefit analysis, relied in part on the propositions that (1) cost-benefit analysis is precluded under the BATEA and BADT tests; and (2) that, insofar as the permissibility of cost-benefit analysis is concerned, the BTA test (the one at issue here) is to be treated the same as those two. See 475 F. 3d, at 98. It is not obvious to us that the first of these *222propositions is correct, but we need not pursue that point, since we assuredly do not agree with the second. It is certainly reasonable for the agency to conclude that the BTA test need not be interpreted to permit only what those other two tests permit. Its text is not identical to theirs. It has the relatively modest goal of “minimizing adverse environmental impact” as compared with the BATE A's goal of “eliminating the discharge of all pollutants.” And it is unencumbered by specified statutory factors of the sort provided for those other two tests, which omission can reasonably be interpreted to suggest that the EPA is accorded greater discretion in determining its precise content.
Respondents and the dissent argue that the mere fact that § 1326(b) does not expressly authorize cost-benefit analysis for the BTA test, though it does so for two of the other tests, displays an intent to forbid its use. This surely proves too much. For while it is true that two of the other tests authorize cost-benefit analysis, it is also true that all four of the other tests expressly authorize some consideration of costs. Thus, if respondents’ and the dissent’s conclusion regarding the import of § 1326(b)’s silence is correct, it is a fortiori true that the BTA test permits no consideration of cost whatsoever, not even the “cost-effectiveness” and “feasibility” analysis that the Second Circuit approved, see supra, at 217, that the dissent would approve, post, at 237, and that respondents acknowledge. The inference that respondents and the dissent would draw from the silence is, in any event, implausible, as § 1326(b) is silent not only with respect to cost-benefit analysis but with respect to all potentially relevant factors. If silence here implies prohibition, then the EPA could not consider any factors in implementing § 1326(b) — an obvious logical impossibility. It is eminently reasonable to conclude that §1326(b)’s silence is meant to convey nothing more than a refusal to tie the agency’s hands as to whether cost-benefit analysis should be used, and if so to what degree.
*223Contrary to the dissent’s suggestion, see post, at 238-240, our decisions in Whitman v. American Trucking Assns., Inc., 531 U. S. 457 (2001), and American Textile Mfrs. Institute, Inc. v. Donovan, 452 U. S. 490 (1981), do not undermine this conclusion. In American Trucking, we held that the text of § 109 of the Clean Air Act, “interpreted in its statutory and historical context . . . , unambiguously bars cost considerations” in setting air quality standards under that provision. 531 U. S., at 471. The relevant “statutory context” included other provisions in the Clean Air Act that expressly authorized consideration of costs, whereas §109 did not. Id., at 467-468. American Trucking thus stands for the rather unremarkable proposition that sometimes statutory silence, when viewed in context, is best interpreted as limiting agency discretion. For the reasons discussed earlier, § 1326(b)’s silence cannot bear that interpretation.
In American Textile, the Court relied in part on a statute’s failure to mention cost-benefit analysis in holding that the relevant agency was not required to engage in cost-benefit analysis in setting certain health and safety standards. 452 U. S., at 510-512. But under Chevron, that an agency is not required to do so does not mean that an agency is not permitted to do so.
This extended consideration of the text of § 1326(b), and comparison of that with the text and statutory factors applicable to four parallel provisions of the Clean Water Act, lead us to the conclusion that it was well within the bounds of reasonable interpretation for the EPA to conclude that cost-benefit analysis is not categorically forbidden. Other arguments may be available to preclude such a rigorous form of cost-benefit analysis as that which was prescribed under the statute’s former BPT standard, which required weighing “the total cost of application of technology” against “the ... benefits to be achieved.” See supra, at 221. But that question is not before us.
*224In the Phase II requirements challenged here the EPA sought only to avoid extreme disparities between costs and benefits. The agency limited variances from the Phase II “national performance standards” to circumstances where the costs are “significantly greater than the benefits” of compliance. 40 CFR § 125.94(a)(5)(h). In defining the “national performance standards” themselves the EPA assumed the application of technologies whose benefits “approach those estimated” for closed-cycle cooling systems at a fraction of the cost: $889 million per year, 69 Fed. Reg. 41666, as compared with (1) at least $3.5 billion per year to operate compliant closed-cycle cooling systems, id., at 41605 (or $1 billion per year to impose similar requirements on a subset of Phase II facilities, id., at 41606), and (2) significant reduction in the energy output of the altered facilities, id., at 41605. And finally, the EPA’s assessment of the relatively meager financial benefits of the Phase II regulations that it adopted— reduced impingement and entrainment of 1.4 billion aquatic organisms, id., at 41661, Exh. XII-6, with annualized use benefits of $83 million, id., at 41662, and nonuse benefits of indeterminate value, id., at 41660-41661 — when compared to annual costs of $389 million, demonstrates quite clearly that the agency did not select the Phase II regulatory requirements because their benefits equaled their costs.
While not conclusive, it surely tends to show that the EPA’s current practice is a reasonable and hence legitimate exercise of its discretion to weigh benefits against costs that the agency has been proceeding in essentially this fashion for over 30 years. See Alaska Dept. of Environmental Conservation v. EPA, 540 U. S. 461, 487 (2004); Barnhart v. Walton, 535 U. S. 212, 219-220 (2002). As early as 1977, the agency determined that, while § 1326(b) does not require cost-benefit analysis, it is also not reasonable to “interpret Section [1326(b)] as requiring use of technology whose cost is wholly disproportionate to the environmental benefit to be gained.” In re Public Service Co. of New Hampshire, 1 *225E. A. D. 332, 340 (1977). See also In re Central Hudson Gas and Electric Corp., EPA General Counsel Opinions, NPDES Permits, No. 63, pp. 371, 381 (July 29, 1977) (“EPA ultimately must demonstrate that the present value of the cumulative annual cost of modifications to cooling water intake structures is not wholly out of proportion to the magnitude of the estimated environmental gains”); Seacoast Anti-Pollution League v. Costle, 597 F. 2d 306, 311 (CA1 1979) (rejecting challenge to an EPA permit decision that was based in part on the agency’s determination that further restrictions would be “‘wholly disproportionate to any environmental benefit’ ”). While the EPA’s prior “wholly disproportionate” standard may be somewhat different from its current “significantly greater than” standard, there is nothing in the statute that would indicate that the former is a permissible interpretation while the latter is not.
Indeed, in its review of the EPA’s Phase I regulations, the Second Circuit seemed to recognize that § 1326(b) permits some form of cost-benefit analysis. In considering a challenge to the EPA’s rejection of dry cooling systems7 as the “best technology available” for Phase I facilities, the Second Circuit noted that “while it certainly sounds substantial that dry cooling is 95 percent more effective than closed-cycle cooling, it is undeniably relevant that that difference represents a relatively small improvement over closed-cycle cooling at a very significant cost.” Riverkeeper, 358 F. 3d, at 194, n. 22. And in the decision below rejecting the use of cost-benefit analysis in the Phase II regulations, the Second Circuit nonetheless interpreted “best technology available” as mandating only those technologies that can “be reasonably borne by the industry.” 475 F. 3d, at 99. But whether it is “reasonable” to bear a particular cost may well depend on *226the resulting benefits; if the only relevant factor was the feasibility of the costs, their reasonableness would be irrelevant.
In the last analysis, even respondents ultimately recognize that some form of cost-benefit analysis is permissible. They acknowledge that the statute’s language is “plainly not so constricted as to require EPA to require industry petitioners to spend billions to save one more fish or plankton.” Brief for Respondent Riverkeeper, Inc., et al. 29. This concedes the principle — the permissibility of at least some cost-benefit analysis — and we see no statutory basis for limiting its use to situations where the benefits are de minimis rather than significantly disproportionate.
* * *
We conclude that the EPA permissibly relied on cost-benefit analysis in setting the national performance standards and in providing for cost-benefit variances from those standards as part of the Phase II regulations. The Court of Appeals’ reliance in part on the agency’s use of cost-benefit analysis in invalidating the site-specific cost-benefit variance provision, 475 F. 3d, at 114, was therefore in error, as was its remand of the national performance standards for clarification of whether cost-benefit analysis was impermissibly used, id., at 104-105. We of course express no view on the remaining bases for the Second Circuit’s remand which did not depend on the permissibility of cost-benefit analysis. See id., at 108, 110, 113, 115, 117, 120.8 The judgment of the *227Court of Appeals is reversed, and the cases are remanded for further proceedings consistent with this opinion.
It is so ordered.
APPENDIX
[[Image here]]
*228[[Image here]]
*229[[Image here]]
The EPA and its Administrator appeared as respondents in support of petitioners. See Brief for Federal Parties as Respondents Supporting Petitioners. References to “respondents” throughout the opinion refer only to those parties challenging the EPA rules at issue in these cases.
Closed-cycle cooling systems recirculate the water used to cool the facility, and consequently extract less water from the adjacent waterway, proportionately reducing impingement and entrainment. Riverkeeper, Inc. v. EPA, 358 F. Bd 174, 182, n. 5 (CA2 2004); 69 Fed. Reg. 41601, and n. 44 (2004).
The EPA has also adopted Phase III rules for facilities not subject to the Phase I and Phase II regulations. 71 Fed. Reg. 35006 (2006). A challenge to those regulations is currently before the Fifth Circuit, where proceedings have been stayed pending disposition of these cases. See ConocoPhillips Co. v. EPA, No. 06-60662.
The dissent finds it “puzzling” that we invoke this proposition (that a reasonable agency interpretation prevails) at the “outset,” omitting the supposedly prior inquiry of‘“whether Congress has directly spoken to the precise question at issue.’” Post, at 241, n. 5 (opinion of Stevens, J.) (quoting Chevron, 467 U. S., at 842). But surely if Congress has directly spoken to an issue then any agency interpretation contradicting what Congress has said would be unreasonable.
What is truly “puzzling” is the dissent’s accompanying charge that the Court’s failure to conduct the Chevron step-one inquiry at the outset “reflects [its] reluctance to consider the possibility ... that Congress’ silence may have meant to foreclose cost-benefit analysis.” Post, at 241, n. 5. Our discussion of that issue, infra, at 222-223, speaks for itself
Respondents concede that the term “available” is ambiguous, as it could mean either technologically feasible or economically feasible. But any ambiguity in the term “available” is largely irrelevant. Regardless of the criteria that render a technology “available,” the EPA would still have to determine which available technology is the “best” one. And as discussed above, that determination may well involve consideration of the technology’s relative costs and benefits.
The statute does not contain a hyphen between the words “conventional” and “pollutant.” “Conventional pollutant” is a statutory term, however, see 33 U. S. C. § 1314(a)(4), and it is clear that in § 1311(b)(2)(E) the adjective modifies “pollutant” rather than “control technology.” The hyphen makes that clear.
Dry cooling systems use air drafts to remove heat, and accordingly remove little or no water from surrounding water sources. See 66 Fed. Reg. 65282 (2001).
Justice Breyer would remand for the additional reason of what he regards as the agency’s inadequate explanation of the change in its criterion for variances — from a relationship of costs to benefits that is “ ‘wholly disproportionate’” to one that is “‘significantly greater.’” Post, at 236 (opinion concurring in part and dissenting in part). That question can have no bearing upon whether the EPA can use cost-benefit analysis, which is the only question presented here. It seems to us, in any case, that the EPA’s explanation was ample. It explained that the “wholly out of proportion” standard was inappropriate for the existing facilities subject to the Phase II rules because those facilities lack “the greater flexibil*227ity available to new facilities for selecting the location of their intakes and installing technologies at lower costs relative to the costs associated with retrofitting existing facilities,” and because “economically impracticable impacts on energy prices, production costs, and energy production . . . could occur if large numbers of Phase II existing facilities incurred costs that were more than ‘significantly greater’ than but not ‘wholly out of proportion’ to the costs in the EPA’s record.” 68 Fed. Reg. 13541 (2003).