delivered the opinion of the Court.

In the 1970s, Congress added two air pollution control schemes to the Clean Air Act: New Source Performance *566Standards (NSPS) and Prevention of Significant Deterioration (PSD), each of them covering modified, as well as new, stationary sources of air pollution. The NSPS provisions define the term “modification,” 42 U. S. C. § 7411(a)(4), while the PSD provisions use that word “as defined in” NSPS, § 7479(2)(C). The Court of Appeals concluded that the statute requires the Environmental Protection Agency (EPA) to conform its PSD regulations on “modification” to their NSPS counterparts, and that EPA’s 1980 PSD regulations can be given this conforming construction. We hold that the Court of Appeals’s reading of the 1980 PSD regulations, intended to align them with NSPS, was inconsistent with their terms and effectively invalidated them; any such result must be shown to comport with the Act’s restrictions on judicial review of EPA regulations for validity.

I

The Clean Air Amendments of 1970, 84 Stat. 1676, broadened federal authority to combat air pollution, see Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 845-846 (1984), and directed EPA to devise National Ambient Air Quality Standards (NAAQS) limiting various pollutants, which the States were obliged to implement and enforce, 42 U. S. C. §§ 7409, 7410. The amendments dealing with NSPS authorized EPA to require operators of stationary sources of air pollutants to use the best technology for limiting pollution, Chevron, supra, at 846; see also 1 F. Grad, Environmental Law §2.03 [14], p. 2-356 *567(2006), both in newly constructed sources and those undergoing “modification,” 42 U. S. C. § 7411(a)(2). Section 111(a) of the 1970 amendments defined this term within the NSPS scheme as “any physical change in, or change in the method of operation of, a stationary source which increases the amount of any air pollutant emitted by such source or which results in the emission of any air pollutant not previously emitted,” 42 U. S. C. § 7411(a)(4).

EPA’s 1975 regulations implementing NSPS provided generally that “any physical or operational change to an existing facility which results in an increase in the emission rate to the atmosphere of any pollutant to which a standard applies shall be considered a modification within the meaning of section 111.” 40 CFR §60.14(a) (1976). Especially significant here is the identification of an NSPS “modification” as a change that “increase^] . . . the emission rate,” which “shall be expressed as kg/hr of any pollutant discharged into the atmosphere. ” § 60.14(b).1

NSPS, however, did too little to “achiev[e] the ambitious goals of the 1970 Amendments,” R. Belden, Clean Air Act 7 (2001) (hereinafter Belden), and the Clean Air Act Amendments of 1977, 91 Stat. 685, included the PSD provisions, which aimed at giving added protection to air quality in certain parts of the country “notwithstanding attainment and *568maintenance of” the NAAQS. 42 U. S. C. § 7470(1).2 The 1977 amendments required a PSD permit before a “major emitting facility” could be “constructed” in an area covered by the scheme. § 7475(a). As originally enacted, PSD applied only to newly constructed sources, but soon a technical amendment added the following subparagraph: “The term ‘construction’ when used in connection with any source or facility, includes the modification (as defined in section 111(a)) of any source or facility.” §14(a)(54), 91 Stat. 1402, 42 U. S. C. § 7479(2)(C); see also New York v. EPA, 413 F. 3d 3, 13 (CADC 2005) (per curiam). In other words, the “construction” requiring a PSD permit under the statute was made to include (though it was not limited to) a “modification” as defined in the statutory NSPS provisions.

In 1980, EPA issued PSD regulations,3 which “limited the application of [PSD] review” of modified sources to instances of “‘major’ modification,” Belden 46, defined as “any physical change in or change in the method of operation of a major stationary source that would result in a significant net emissions increase of any pollutant subject to regulation under the Act.” 40 CFR § 51.166(b)(2)(i) (1987). Further regulations in turn addressed various elements of this definition, three of which are to the point here. First, the regulations specified that an operational change consisting merely of “[a]n increase in the hours of operation or in the production rate” would not generally constitute a “physical change in or change in the method of operation.” § 51.166(b)(2)(iii)(/). For purposes of a PSD permit, that is, such an operational *569change would not amount to a “modification” as the Act defines it. Second, the PSD regulations defined a “net emissions increase” as “[a]ny increase in actual emissions from a particular physical change or change in the method of operation,” net of other contemporaneous “increases and decreases in actual emissions at the source.” §51.166(b)(3)(i). “Actual emissions” were defined to “equal the average rate, in tons per year, at which the unit actually emitted the pollutant during a two-year period which precedes the particular date and which is representative of normal source operation.” §51.166(b)(21)(ii). “[AJetual emissions” were to be “calculated using the unit’s actual operating hours [and] production rates.” Ibid. Third, the term “significant” was defined as “a rate of emissions that would equal or exceed” one or another enumerated threshold, each expressed in “tons per year.” § 51.166(b)(23)(i).

It would be bold to try to synthesize these statutory and regulatory provisions in a concise paragraph, but three points are relatively clear about the regime that covers this case:

(a) The Act defines modification of a stationary source of a pollutant as a physical change to it, or a change in the method of its operation, that increases the amount of a pollutant discharged or emits a new one.

(b) EPA’s NSPS regulations require a source to use the best available pollution-limiting technology only when a modification would increase the rate of discharge of pollutants measured in kilograms per hour.

(c) EPA’s 1980 PSD regulations require a permit for a modification (with the same statutory definition) only when it is a major one and only when it would increase the actual annual emission of a pollutant above the actual average for the two prior years.

The Court of Appeals held that Congress’s provision defining a PSD modification by reference to an NSPS modifica*570tion caught not only the statutory NSPS definition, but also whatever regulatory gloss EPA puts on that definition at any given time (for the purposes of the best technology requirement). When, therefore, EPA’s PSD regulations specify the “change” that amounts to a “major modification” requiring a PSD permit, they must measure an increase in “the amount of any air pollutant emitted,” 42 U. S. C. § 7411(a)(4), in terms of the hourly rate of discharge, just the way NSPS regulations do. Petitioners and the United States say, on the contrary, that when EPA addresses the object of the PSD scheme it is free to put a different regulatory interpretation on the common statutory core of “modification,” by measuring increased emission not in terms of hourly rate but by the actual, annual discharge of a pollutant that will follow the modification, regardless of rate per hour. This disagreement is the nub of the case.

II

Respondent Duke Energy Corporation runs 30 coal-fired electric generating units at eight plants in North and South Carolina. United States v. Duke Energy Corp., 411 F. 3d 539, 544 (CA4 2005). The units were placed in service between 1940 and 1975, and each includes a boiler containing thousands of steel tubes arranged in sets. Ibid. Between 1988 and 2000,4 Duke replaced or redesigned 29 tube assemblies in order to extend the life of the units and allow them to run longer each day. Ibid.

*571The United States filed this action in 2000, claiming, among other things, that Duke violated the PSD provisions by doing this work without permits. Environmental Defense, North Carolina Sierra Club, and North Carolina Public Interest Research Group Citizen Lobby/Education Fund intervened as plaintiffs and filed a complaint charging similar violations.

Duke moved for summary judgment, one of its positions being that none of the projects was a “major modification” requiring a PSD permit because none increased hourly rates of emissions. The District Court agreed with Duke’s reading of the 1980 PSD regulations. It reasoned that their express exclusion of “ ‘[a]n increase in the hours of operation’ ” from the definition of a “ ‘physical change or change in the method of operation’” implied that “post-project emissions levels must be calculated assuming” preproject hours of operation. 278 F. Supp. 2d 619, 640-641 (MDNC 2003). Consequently, the District Court said, a PSD “major modification” can occur “only if the project increases the hourly rate of emissions.” Id., at 641. The District Court found further support for its construction of the 1980 PSD regulations in one letter and one memorandum written in 1981 by EPA’s Director of the Division of Stationary Source Enforcement, Edward E. Reich. Id., at 641-642.

The United States and intervenor-plaintiffs (collectively, plaintiffs) subsequently stipulated “that they do not contend that the projects at issue in this case caused an increase in the maximum hourly rate of emissions at any of Duke Energy’s units.” App. 504. Rather, their claim “is based solely on their contention that the projects would have been projected to result in an increased utilization of the units at issue.” Ibid. Duke, for its part, stipulated to plaintiffs’ right to appeal the District Court’s determination that projects resulting in greater operating hours are not “major modifications” triggering the PSD permit requirement, absent an increase in the hourly rate of emissions. The Dis*572trict Court then entered summary judgment for Duke on all PSD claims.

The Court of Appeals for the Fourth Circuit affirmed, “albeit for somewhat different reasons.” 411 F. 3d, at 542. “[T]he language and various interpretations of the PSD regulations . . . are largely irrelevant to the proper analysis of this case,” reasoned the Court of Appeals, “because Congress’ decision to create identical statutory definitions of the term ‘modification’” in the NSPS and PSD provisions of the Clean Air Act “has affirmatively mandated that this term be interpreted identically” in the regulations promulgated under those provisions. Id., at 547, n. 3, 550. The Court of Appeals relied principally on the authority of Rowan Cos. v. United States, 452 U. S. 247, 250 (1981), where we held against the Government’s differing interpretations of the word “wages” in different tax provisions. 411 F. 3d, at 550. As the Court of Appeals saw it, Rowan establishes an “effectively irrebuttable” presumption that PSD regulations must contain the same conditions for a “modification” as the NSPS regulations, including an increase in the hourly rate of emissions.5 411 F. 3d, at 550.

As the Court of Appeals said, Duke had not initially relied on Rowan, see 411 F. 3d, at 547, n. 4, and when the Court sua sponte requested supplemental briefing on Rowan's relevance, plaintiffs injected a new issue into the case. They argued that a claim that the 1980 PSD regulation exceeded statutory authority would be an attack on the validity of the regulation that could not be raised in an enforcement proceeding. See 42 U. S. C. § 7607(b)(2). Under § 307(b) of the *573Act, they said, judicial review for validity can be obtained only by a petition to the Court of Appeals for the District of Columbia Circuit, generally within 60 days of EPA’s rule-making. 42 U. S. C. § 7607(b).

The Court of Appeals rejected this argument. “Our choice of this interpretation of the PSD regulations ... is not an invalidation of those regulations,” it said, because “the PSD regulations can be interpreted” to require an increase in the hourly emissions rate as an element of a major “modification” triggering the permit requirement. 411 F. 3d, at 549, n. 7. To show that the 1980 PSD regulations are open to this construction, the Court of Appeals cited the conclusions of the District Court and the Reich opinions.

We granted the petition for certiorari brought by intervenor-plaintiffs, 547 U. S. 1127 (2006), and now vacate.

III

The Court of Appeals understood that it was simply construing EPA’s 1980 PSD regulations in a permissible way that left them in harmony with their NSPS counterpart and, hence, the Act’s single definition of “modification.” The plaintiffs say that the Court of Appeals was rewriting the PSD regulations in a way neither required by the Act nor consistent with their own text.

It is true that no precise line runs between a purposeful but permissible reading of the regulation adopted to bring it into harmony with the Court of Appeals’s view of the statute, and a determination that the regulation as written is invalid. But the latter occurred here, for the Court of Appeals’s efforts to trim the PSD regulations to match their different NSPS counterparts can only be seen as an implicit declaration that the PSD regulations were invalid as written.

A

In applying the 1980 PSD regulations to Duke’s conduct, the Court of Appeals thought that, by defining the term *574“modification” identically in its NSPS and PSD provisions, the Act required EPA to conform its PSD interpretation of that definition to any such interpretation it reasonably adhered to under NSPS. But principles of statutory construction are not so rigid. Although we presume that the same term has the same meaning when it occurs here and there in a single statute, the Court of Appeals mischaracterized that presumption as “effectively irrebuttable.” 411 F. 3d, at 550. We also understand that “[m]ost words have different shades of meaning and consequently may be variously construed, not only when they occur in different statutes, but when used more than once in the same statute or even in the same section.” Atlantic Cleaners & Dyers, Inc. v. United States, 286 U. S. 427, 433 (1932). Thus, the “natural presumption that identical words used in different parts of the same act are intended to have the same meaning ... is not rigid and readily yields whenever there is such variation in the connection in which the words are used as reasonably to warrant the conclusion that they were employed in different parts of the act with different intent.” Ibid. A given term in the same statute may take on distinct characters from association with distinct statutory objects calling for different implementation strategies.

The point is the same even when the terms share a common statutory definition, if it is general enough, as we recognized in Robinson v. Shell Oil Co., 519 U. S. 337 (1997). There the question was whether the term “employees” in § 704(a) of Title VII of the Civil Rights Act of 1964 covered former employees. Title VII expressly defined the term “employee,” 42 U. S. C. § 2000e(f), but the definition was “consistent with either current or past employment,” 519 U. S., at 342, and we held that “each section” of Title VII “must be analyzed to determine whether the context gives the term a further meaning that would- resolve the issue in dispute,” id., at 343-344.

*575If Robinson were inconsistent with Rowan (on which the Court of Appeals relied), it would be significant that Robinson is the later case, but we read the two as compatible. In Rowan, the question was whether the value of meals and lodging given to employees by an employer for its own convenience should be counted in computing “wages” under the Federal Insurance Contributions Act (FICA), 26 U. S. C. § 3101 et seq. (2000 ed. and Supp. IV), and the Federal Unemployment Tax Act (FUTA), 26 U. S. C. § 3301 et seq. (2000 ed. and Supp. IV). Treasury Regulations made this value “includable in ‘wages’ as defined in FICA and FUTA, even though excludable from ‘wages’ under the substantially identical” statutory definition of “wages” for income-tax withholding purposes. 452 U. S., at 252. Although we ultimately held that the income-tax treatment was the proper one across the board, we did not see it this way simply because a “substantially identical” definition of “wages” appeared in each of the different statutory provisions. Instead, we relied on a manifest “congressional concern for the interest of simplicity and ease of administration.” Id., at 255 (internal quotation marks omitted). The FICA and FUTA regulations fell for failing to “serve that interest,” id., at 257, not for defying definitional identity.

In fact, in a setting much like Rowan, we recently declined to require uniformity when resolving ambiguities in identical statutory terms. In United States v. Cleveland Indians Baseball Co., 532 U. S. 200 (2001), we rejected the notion that using the phrase “wages paid” in both “the discrete taxation and benefits eligibility contexts” can, standing alone, “compel symmetrical construction,” id., at 213; we gave “substantial judicial deference” to the “longstanding,” “reasonable,” and differing interpretations adopted by the Internal Revenue Service in its regulations and Revenue Rulings. Id., at 218-220. There is, then, no “effectively irrebuttable” presumption that the same defined term in different provisions of the *576same statute must “be interpreted identically.” 411 F. 3d, at 550. Context counts.

It is true that the Clean Air Act did not merely repeat the term “modification” or the same definition of that word in its NSPS and PSD sections; the PSD language referred back to the section defining “modification” for NSPS purposes. 42 U. S. C. §7479(2)(C).But that did not matter in Robinson, and we do not see the distinction as making any difference here. Nothing in the text or the legislative history of the technical amendments that added the cross-reference to NSPS suggests that Congress had details of regulatory implementation in mind when it imposed PSD requirements on modified sources; the cross-reference alone is certainly no unambiguous congressional code for eliminating the customary agency discretion to resolve questions about a statutory definition by looking to the surroundings of the defined term, where it occurs. See New York, 413 F. 3d, at 19 (“So far as appears,... [this] incorporation by reference [is] the equivalent of Congress’s having simply repeated in the [PSD] context the definitional language used before in the NSPS context”); cf. 91 Stat. 745 (expressly incorporating in an unrelated provision of the 1977 amendments “the interpretative regulation of the [EPA] Administrator . .. published in 41 Federal Register 55524-30” with specified exceptions); New York, supra, at 19 (“Congress’s failure to use such an express incorporation of prior regulations for ‘modification’ cuts against” any suggestion that “Congress intended to incorporate” into the Act the “preexisting regulatory definition” of “modification”). Absent any iron rule to ignore the reasons for regulating PSD and NSPS “modifications” differently, EPA’s construction need do no more than fall within the limits of what is reasonable, as set by the Act’s common 6 definition.

*577B

The Court of Appeals’s reasoning that the PSD regulations must conform to their NSPS counterparts led the court to read those PSD regulations in a way that seems to us too far a stretch for the language used. The 1980 PSD regulations on “modification” simply cannot be taken to track the Agency’s regulatory definition under the NSPS.

True, the 1980 PSD regulations may be no seamless narrative, but they clearly do not define a “major modification” in terms of an increase in the “hourly emissions rate.” On its face, the definition in the PSD regulations specifies no rate at all, hourly or annual, merely requiring a physical or operational change “that would result in a significant net emissions increase of any” regulated pollutant. 40 CFR § 51.166(b)(2)(i). But even when a rate is mentioned, as in the regulatory definitions of the two terms, “significant” and “net emissions increase,” the rate is annual, not hourly. Each of the thresholds that quantify “significant” is described in “tons per year,” § 51.166(b)(23)(i), and a “net emissions increase” is an “increase in actual emissions” measured against an “average” prior emissions rate of so many “tons *578per year,” §§51.166(b)(3)(i) and (21)(ii). And what is further at odds with the idea that hourly rate is relevant is the mandate that “[ajctual emissions shall be calculated using the unit’s actual operating hours,” § 51.166(b)(21)(ii), since “actual emissions” must be measured in a manner that looks to the number of hours the unit is or probably will be actually running. What these provisions are getting at is a measure of actual operations averaged over time, and the regulatory language simply cannot be squared with a regime under which “hourly rate of emissions,” 411 F. 3d, at 550 (emphasis deleted), is dispositive.

The reasons invoked by the Court of Appeals for its different view are no match for these textual differences. The appellate court cited two authorities ostensibly demonstrating that the 1980 PSD regulations “can be interpreted consistently” with the hourly emissions test, the first being the analysis of the District Court in this case. Id., at 549, n. 7. The District Court thought that an increase in the hourly emissions rate was necessarily a prerequisite to a PSD “major modification” because a provision of the 1980 PSD regulations excluded an “ ‘increase in the hours of operation or in the production rate’ ” from the scope of “ ‘[a] physical change or change in the method of operation.’ ” 278 F. Supp. 2d, at 640-641 (quoting 40 CFR §§51.166(b)(2)(iii)(/) and (3)(i)(a) (1987)). The District Court read this exclusion to require, in effect, that a source’s hours of operation “be held constant” when preproject emissions are being compared with postproject emissions for the purpose of calculating the “net emissions increase.” 278 F. Supp. 2d, at 640.

We think this understanding of the 1980 PSD regulations makes the mistake of overlooking the difference between the two separate components of the regulatory definition of “major modification”: “[1] any physical change in or change in the method of operation of a major stationary source that [2] would result in a significant net emissions increase *579of any pollutant subject to regulation under the Act.” § 51.166(b)(2)(i); cf. New York, 413 F. 3d, at 11 (“[The statutory] definition requires both a change — whether physical or operational — and a resulting increase in emissions of a pollutant” (emphasis in original)); Wisconsin Elec. Power Co. v. Reilly, 893 F. 2d 901, 907 (CA7 1990) (same). The exclusion of “increase in . . . hours . . . or . . . production rate,” §51.166(b)(2)(iii)(/), speaks to the first of these components (“physical change ... or change in . . . method,” § 51.166(b)(2)(i)), but not to the second (“significant net emissions increase,” ibid.). As the preamble to the 1980 PSD regulations explains, forcing companies to obtain a PSD permit before they could simply adjust operating hours “would severely and unduly hamper the ability of any company to take advantage of favorable market conditions.” 45 Fed. Reg. 52704. In other words, a mere increase in the hours of operation, standing alone, is not a “physical change or change in the method of operation.” 40 CFR §51.166(b)(2)(iii).

But the District Court took this language a step further. It assumed that increases in operating hours (resulting in emissions increases at the old rate per hour) must be ignored even if caused or enabled by an independent “physical change ... or change in the method of operation.” §51.166(b)(2)(i). That reading, however, turns an exception to the first component of the definition into a mandate to ignore the very facts that would count under the second, which defines “net emissions increase” in terms of “actual emissions,” §51.166(b)(3)(i), during “the unit’s actual operating hours,” § 51.166(b)(21)(ii); see also 57 Fed. Reg. 32328 (1992) (“[A]n increase in emissions attributable to an increase in hours of operation or production rate which is the result of a construction-related activity is not excluded from [PSD] review .. . ”).7

*580The Court of Appeals invoked one other source of support, the suggestion in the Reich opinions that a physical or operational change increasing a source’s hours of operation, without an increase in the hourly emissions rate, cannot be a PSD “major modification.” Duke continues to rely on those opinions here, asserting that “there are no contrary Agency pronouncements.” Brief for Respondent Duke 28. The Reich letters are not, however, heavy ammunition. Their persuasiveness is elusive, neither of them containing more than one brief and conelusory statement supporting Duke’s position. Nor, it seems, are they unembarrassed by any “contrary Agency pronouncements.” See, e.g., App. 258 (Memorandum of Don R. Clay, Acting Assistant EPA Administrator for Air and Radiation (Sept. 9,1988) (when “plans to increase production rate or hours of operation are inextricably intertwined with the physical changes planned,” they are “precisely the type of change in hours or rate o[f] operation that would disturb a prior assessment of a source’s environmental impact and should have to undergo PSD review scrutiny” (internal quotation marks and alterations omitted))); see also 57 Fed. Reg. 32328. In any event, it answers the citation of the Reich letters to realize that an isolated opinion of *581an agency official does not authorize a court to read a regulation inconsistently with its language.8

In sum, the text of the 1980 PSD regulations on “modification” doomed the Court of Appeals’s attempt to equate those regulations with their NSPS counterpart. As a consequence, we have to see the Court of Appeals’s construction of the 1980 PSD regulations as an implicit invalidation of those regulations, a form of judicial review implicating the provisions of § 307(b) of the Act, which limit challenges to the validity of a regulation during enforcement proceedings when such review “could have been obtained” in the Court of Appeals for the District of Columbia Circuit within 60 days of EPA rulemaking. See 42 U. S. C. § 7607(b); see also United States v. Cinergy Corp., 458 F. 3d 705, 707-708 (CA7 2006); Wisconsin Elec. Power Co., 893 F. 2d, at 914, n. 6. Because the Court of Appeals did not believe that its analysis reached validity, it did not consider the applicability or effect of that limitation here. We have no occasion at this point to consider the significance of § 307(b) ourselves.

IV

Finally, Duke assumes for argument that the Act and the 1980 regulations may authorize EPA to construe a PSD “modification” as it has done, but it charges that the agency has taken inconsistent positions and is now “retroactively *582targeting twenty years of accepted practice.” Brief for Respondent Duke 37; see also Brief for State of Alabama et al. as Amici Curiae. This claim, too, has not been tackled by the District Court or the Court of Appeals; to the extent it is not procedurally foreclosed, Duke may press it on remand.

* * *

The judgment of the Court of Appeals is vacated, and the case is remanded for further proceedings consistent with this opinion.

It is so ordered.

concurring in part.

I join all but Part III-A of the Court’s opinion. I write separately to note my disagreement with the dicta in that portion of the opinion, which states that the statutory cross-reference does not mandate a singular regulatory construction.

The Prevention of Significant Deterioration (PSD) statute explicitly links the definition of the term “modification” to that term’s definition in the New Source Performance Standard (NSPS) statute:

“The term ‘construction’ when used in connection with any source or facility, includes the modification (as defined in section 7411(a) of this title) of any source or facility.” 42 U. S. C. § 7479(2)(C).

Section 7411(a) contains the NSPS definition of “modification,” which the parties agree is the relevant statutory definition of the term for both PSD and NSPS. Because of the cross-reference, the definitions of “modification” in PSD and NSPS are one and the same. The term “modification” therefore has the same meaning despite contextual variations in the two admittedly different statutory schemes. Congress’ explicit linkage of PSD’s definition of “modification” to NSPS’ prevents the Environmental Protection *583Agency (EPA) from adopting differing regulatory definitions of “modification” for PSD and NSPS. Cf. IBP, Inc. v. Alvarez, 546 U. S. 21, 34 (2005) (concluding that an “explicit reference” to a previous statutory definition prohibits interpreting the same word differently).

Section 7479(2)(C)’s cross-reference carries more meaning than the mere repetition of the same word in a different statutory context. When Congress repeats the same word in a different statutory context, it is possible that Congress might have intended the context to alter the meaning of the word. See Atlantic Cleaners & Dyers, Inc. v. United States, 286 U. S. 427, 433 (1932). No such possibility exists with §7479(2)(C). By incorporating NSPS’ definition of “modification,” Congress demonstrated that it did not intend for PSD’s definition of “modification” to hinge on contextual factors unique to the PSD statutory scheme. Thus, United States v. Cleveland Indians Baseball Co., 532 U. S. 200 (2001), which analyzes the mere repetition of the same word in a different statutory context, carries little weight in this situation.

Likewise, this case differs from the circumstance we faced in Robinson v. Shell Oil Co., 519 U. S. 337 (1997). In Robinson, we considered whether “employee,” as used in § 704(a) of Title VII of the Civil Rights Act of 1964, included former employees. We determined that under the clear language of the statute, certain statutory provisions using the term “employee” made sense only with respect to former employees or current employees, but not both. Id., at 342-343. Accordingly, upon analyzing the context of § 704(a), we were compelled to conclude that the term “employee” included former employees. This case does not present a similar situation. The statute here includes a statutory cross-reference, which conveys a clear congressional intent to provide a common definition for the term “modification.” And the contextual differences between PSD and NSPS do not compel *584different meanings for the term “modification.” Robinson is, therefore, inapplicable.

Even if the cross-reference were merely the equivalent of repeating the words of the definition, we must still apply our usual presumption that the same words repeated in different parts of the same statute have the same meaning. See Atlantic Cleaners, supra, at 483; ante, at 574. That presumption has not been overcome here. While the broadly stated regulatory goals of PSD and NSPS differ, these contextual differences do not compel different definitions of “modification.” That is, unlike in Robinson, reading the statutory definition in the separate contexts of PSD and NSPS does not require different interpretations of the term “modification.” EPA demonstrated as much when it recently proposed regulations that would unify the regulatory definitions of “modification.” See 70 Fed. Reg. 61083, n. 3 (2005) (terming the proposal “an appropriate exercise of our discretion” and stating that the unified definition better serves PSD’s goals).

The majority opinion does little to overcome the presumption that the same words, when repeated, carry the same meaning. Instead, it explains that this Court’s cases do not compel identical language to be interpreted identically in all situations. Granting that point, the majority still has the burden of stating why our general presumption does not control the outcome here. It has not done so.