dissenting.
The Court today interprets the term “original sentence,” as it appears in 18 U. S. C. § 3565(a), to mean “the maximum sentence, under the relevant Sentencing Guidelines range, which a defendant could have received, but did not, when initially sentenced.” I think this interpretation ignores the *70most natural meaning of these two words, and I therefore dissent.
Section 3565(a) does not indicate on its face whether a defendant found in violation of probation must be sentenced to prison or resentenced to another term of probation. I agree with the Court that § 3565(a) must be read to require imposition of a term of imprisonment; otherwise, as the Court explains, the proviso would be senseless.1 See ante, at 45; In re Chapman, 166 U. S. 661, 667 (1897) (“[Nothing is better settled than that statutes should receive a sensible construction, such as will effectuate the legislative intention, and, if possible, so as to avoid an unjust or an absurd conclusion”). If the Court had stopped there, I would have been happy to join its' opinion. Having correctly resolved one ambiguity in § 3565(a), however, the Court proceeds to find another, regarding the meaning of the term “original sentence,” where none exists. The Court thus ultimately concludes, incorrectly in my view, that the rule of lenity should be applied.
The Court believes that the Government’s reading of §3565(a) is not “unambiguously correct.” Ante, at 54. As we have explained, however, the rule of lenity should not be applied “merely because it [is] possible to articulate a construction more narrow than that urged by the Government.” Moskal v. United States, 498 U. S. 103, 108 (1990). Instead we have reserved lenity for those situations where, after “[a]pplying well-established principles of statutory construction,” Gozlon-Peretz v. United States, 498 U. S. 395, 410 (1991), there still remains “a grievous ambiguity or uncer*71tainty in the language and structure of the Act,” Chapman v. United States, 500 U. S. 453, 463 (1991) (internal quotation marks and citation omitted).
The term “original sentence” is not defined in the statute. A basic principle of statutory construction provides that where words in a statute are not defined, they “must be given their ordinary meaning.” Id., at 462; see also Smith v. United States, 508 U. S. 223, 228 (1993) (“When a word is not defined by statute, we normally construe it in accord with its ordinary or natural meaning”).
Whether one consults a dictionary or common sense, the meaning of “original sentence” is plain: The term refers to the initial judgment imposing punishment on a defendant. “Original” is commonly understood to mean “initial” or “first in order.” See Webster’s Third New International Dictionary 1592 (1971) (Webster’s) (defining “original” as “of or relating to a rise or beginning . . . initial, primary”); Black’s Law Dictionary 1099 (6th ed. 1990) (defining “original” as “[primitive” or “first in order”). “Sentence,” in turn, is ordinarily meant in the context of criminal law to refer to the judgment or order “by which a court or judge imposes punishment or penalty upon a person found guilty.” Webster’s 2068; see also Black’s Law Dictionary, supra, at 1362 (defining “sentence” as “[t]he judgment. . . imposing the punishment to be inflicted, usually in the form of a fine, incarceration, or probation”).2 In the context of § 3565(a), the term “original sentence” thus must refer to the sentence of probation a defendant actually received when initially sentenced. It cannot, therefore, mean what the Court says it means: the maximum sentence which a defendant could have received, but did not.
The Court’s interpretation thus founders, I believe, because the word “sentence” does not ordinarily, or even occa*72sionally, refer to a range of available punishment. Nor does the modifying word “original” support the Court’s interpretation, because “original” is nowhere defined as “potential” or “available,” nor can it be so construed. Yet under the Court’s interpretation of the term “original sentence,” if we know that “sentence” itself does not mean an available range of punishment, then “original” must be twisted to mean what we know it cannot — i. e., “potential” or “available.”3
This Court has on many occasions demonstrated its clear understanding of the term “original sentence.” See, e. g., Hicks v. Feiock, 485 U. S. 624, 639, and n. 11 (1988) (using term “original sentence” to refer to sentence of imprisonment initially imposed and suspended); Tuten v. United States, 460 U. S. 660, 666-667, and n. 11 (1983) (using term “original sentence” to refer to period of probation imposed by sentencing court when youthful defendant was initially sentenced); United States v. DiFrancesco, 449 U. S. 117, 135 (1980), and id., at 148 (Brennan, J., dissenting) (both using term “original sentence” to refer to sentence imposed upon defendant at conclusion of first trial); North Carolina v. Pearce, 395 U. S. 711, 713, and n. 1 (1969), and id., at 743 (Black, J., concurring in part and dissenting in part) (same); Pennsylvania ex rel. Sullivan v. Ashe, 302 U. S. 51, 53 (1937) *73(same). As these and numerous other opinions show,4 we have until today invariably used “original sentence” just as one would expect: to refer to the punishment imposed upon a defendant when he was first sentenced, and to distinguish that initial sentence from a sentence the defendant received after some intervening event — such as a new trial, see Pearce, swpra, or a revocation of probation, see Hicks, supra5
The Court’s heretofore firm grasp on the meaning of “original sentence” should not be cause for wonder or surprise. Whether alone or in combination, the definitions of “original” and “sentence” simply do not seem open to serious debate. Once the term “original sentence” is accorded its ordinary meaning, the operation of § 3565(a) becomes perfectly clear.6 *74It follows, from another elementary canon of construction, that the plain language of § 3565(a) should control. See Moskal, 498 U. S., at 108. As we stated in Consumer Product Safety Comm’n v. GTE Sylvania, Inc., 447 U. S. 102, 108 (1980), “[a]bsent a clearly expressed legislative intention to the contrary, [the statutory] language must ordinarily be regarded as conclusive.”7
The Court offers several reasons for rejecting the most natural reading of § 3565(a). None of them persuades. The Court begins by suggesting that if Congress meant for the sentence of probation to be used to calculate the length of incarceration, it could have stated so more clearly. See ante, at 46. Although perhaps true, Congress could have just as easily, if it wished, stated in clear terms that the sentence of incarceration should be calculated based on the maximum available sentence under the Guidelines range. Indeed, as I have already noted, supra, at 72, n. 3, Congress stated something very similar in the subsections preceding and following the one at issue, where it provided that upon revocation of probation, a court can or must impose any sentence that was “available” when the defendant was initially sentenced. See §§ 3565(a)(2) and (b); United States v. Sosa, 997 F. 2d 1130, 1133 (CA5 1993); United States v. Byrkett, 961 F. 2d 1399, 1400-1401 (CA8 1992) (“If Congress, in referring to the ‘original sentence,’ meant the Guidelines range *75applicable at the time of the initial sentencing, it would have simply said, ‘any other sentence that was available ... at the time of the initial sentencing,’ as it did” in §§ 3565(a)(2) and (b)).
The Court also asserts that its reading of the term avoids according two different meanings to the word “sentence.” Yet under the Court’s own interpretation, the word “sentence” when used as a verb refers to the imposition of a fixed period of incarceration; but when the word “sentence” next appears, as a noun, the Court concludes that it refers to a range of available punishment. Thus it is the Court’s reading of the statute that fails “‘to give ... a similar construction’ ” to a word used as both a noun and a verb in a single statutory sentence. See ante, at 46 (quoting Reves v. Ernst & Young, 507 U. S. 170, 177 (1993)). Under what I think is the correct reading of the statute, all that changes is what the defendant will be (or was) sentenced to — prison or probation; the word “sentence” itself does not change meanings.
The Court next contends that “ ‘[probation and imprisonment are not fungible,’” ante, at 46 (citation omitted), and that its interpretation of the statute avoids the “shoal” supposedly encountered when explaining “how multiplying a sentence of probation by one-third can yield a sentence of imprisonment,” ante, at 47. Probation and imprisonment, however, need not be fungible for this statute to make sense. They need only both be subsumed under the term “sentence,” which, for the reasons previously stated, they are. See Black’s Law Dictionary, at 1362 (defining “sentence” as a judgment imposing punishment, which may include “a fine, incarceration, or probation”). While tying the length of imprisonment to the length of the original sentence of probation might seem harsh to the Court, surely it is not an irrational method of calculation. Indeed, the Court does not question that Congress could have tied the length of imprisonment to the length of the original sentence of probation.
*76Congress in fact prescribed a similar method of calculation in a parallel provision of the Anti-Drug Abuse Act of 1988, 18 U. S. C. § 3583(g), which was added at the same time as § 3565(a) and which also sets out the punishment for defendants found in possession of a controlled substance. Section 3583(g) explicitly provides: “If the defendant is found by the court to be in the possession of a controlled substance, the court shall terminate the term of supervised release and require the defendant to serve in prison not less than one-third of the term of supervised release.” Considering that §§ 3565(a) and 3583(g) were enacted at the same time and are directed at precisely the same problem, it seems quite reasonable to construe them in pari materia to call for parallel treatment of drug offenders under noncustodial supervision. Whatever the differences between supervised release and probation, surely supervised release is more like probation than it is like imprisonment. That Congress explicitly chose in § 3583(g) to tie the length of imprisonment to the length of supervised release suggests quite strongly that Congress meant in § 3565(a) to use length of the original sentence of probation as the basis for calculation. At the very least, the method of calculation prescribed in § 3583(g) removes the imaginary “shoal” which blocks the Court’s way to a sensible construction of § 3565(a).
The Court refuses to read these provisions in pari materia because a sentence of probation is normally — but not necessarily — longer than a period of supervised release. See ante, at 50-51, and n. 8. Simply because the end result of the calculation might be different in some cases, however, is not a persuasive reason for refusing to recognize the obvious similarity in the methods of calculation. Nor is it irrational for Congress to have decided that, in general, those defendants who have already been incarcerated should return to prison for a shorter time than those who have served no time in prison.
*77Here, as in other portions of its opinion, the Court expresses concern with the apparent harshness of the result if “original sentence” is interpreted to mean the sentence of probation initially imposed on a defendant.8 In some cases the result may indeed appear harsh. Yet harsh punishment, in itself, is neither a legitimate ground for invalidating a statute nor cause for injecting ambiguity into a statute that is susceptible to principled statutory construction. See Callarian v. United States, 364 U. S. 587, 596 (1961) (“The rule [of lenity] comes into operation at the end of the process of construing what Congress has expressed, not at the beginning as an overriding consideration of being lenient to wrongdoers”). A straightforward reading of § 3565(a) may in some cases call for imposition of severe punishment, but it does not produce “a result so absurd or glaringly unjust, as to raise a reasonable doubt about Congress’ intent.” Chapman, 500 U. S., at 463-464 (internal quotation marks and citations omitted).
The Court’s interpretation of § 3565(a), finally, creates an incurable uncertainty: It offers no sound basis for choosing *78which point in the Guidelines range should serve as the basis for calculating a revocation sentence. After describing the four possible reference points within the range, the Court selects the maximum available sentence. It rejects selecting a point in the middle of the available range, because to do so “would be purely arbitrary.” Ante, at 55. Yet the Court does not explain why choosing the top end of the range is any less arbitrary, or any more “sensible,” than picking a point in the middle of the range. Indeed, the Court’s selection smacks of awarding a consolation prize to the Government simply out of concern that the Government was mistakenly done out of victory in the main event. And choosing the maximum possible sentence under the Guidelines hardly seems consistent with the rule of lenity which the Court purports to apply.9
A straightforward reading of § 3565(a) creates no similar uncertainty. Because I think the language of § 3565(a) is clear, I would apply it. Accordingly, I would reverse the Court of Appeals.
The option of imposing a fine after revocation is also foreclosed. As a matter of common usage, the prepositional phrase following a noun need not be repeated when the noun appears again in the same sentence. Thus, § 3565(a) reads: “[T]he court shall revoke the sentence of probation and sentence the defendant to not less than one-third of the original sentence [of probation](Emphasis added.) “[N]ot less than one-third” of a term of probation is a period of time. A fine cannot follow revocation, then, because a fine is measured in money, not time.
Federal sentencing law also consistently uses the word “sentence” to refer to the punishment actually imposed on a defendant. See, e. g., 18 U. S. C. §§ 3551(b) and (c), 3553(a), (b), (c), and (e), and 3554-3658.
Congress itself, in the subsections preceding and following the provision at issue here, distinguishes between “original” and “available.” Sections 3565(a)(2) and (b) provide that under certain circumstances, a court can or must “revoke the sentence of probation and impose any other sentence that was available... at the time of the initial sentencing.” (Emphasis added.) If “original” and “available” were in fact synonymous, or if “sentence” could mean an available range of punishment, Congress could have simply stated in §§ 3565(a)(2) and (b) that upon revocation of probation, a court can or must “impose the original sentence.” See United States v. Sosa, 997 F. 2d 1130, 1133 (CA5 1993) (“The statute taken as a whole demonstrates that Congress knew how to refer to the sentence the defendant could have received at the time of the initial sentencing. Instead, ... Congress used the term ‘original sentence,’ which plainly refers to the sentence imposed on the defendant for his original crime”).
The term “original sentence” appears in at least 50 prior opinions. Rather than citing them all, suffice it to say that a review of these opinions reveals that the term is not once used to refer to the range of punishment potentially applicable when a defendant was first sentenced.
Although the term “original sentence” does not appear in other provisions of the Federal Criminal Code chapter on sentencing, it does appear in other federal statutes and rules. In each instance, the term refers to the sentence initially imposed upon a defendant. See, e.g., Fed. Rule Crim. Proc. 35(a)(2) (directing sentencing courts to correct sentences upon remand from a court of appeals if, after further sentencing proceedings, “the court determines that the original sentence was incorrect”); 10 U. S. C. §863 (providing that upon rehearing in a court-martial, “no sentence in excess of or more severe than the original sentence may be imposed”). The term is similarly used in the Federal Sentencing Guidelines. See, e.g., United States Sentencing Commission, Guidelines Manual §4A1.2(k) (Nov. 1993) (using term “original sentence” to refer to sentence previously imposed upon defendant); § 7B1.4, comment., n. 4 (same).
The Court suggests that if “original sentence” is given its ordinary meaning, the statute will have to be interpreted to require the absurd result that a revocation sentence be another term of probation. See ante, at 47-48, n. 5. I do not see at all how or why the latter proposition follows from the former. The Court rightly rejects interpreting the statute to require reimposition of probation because that would be a senseless reading, and it would be senseless regardless of what the term “original sentence” means. See ante, at 44-45. It is thus beyond me why the Court *74seems to think that according the term “original sentence” its most natural reading would require it to readopt a reading of the statute that it justifiably discarded as senseless.
The Court suggests that the legislative history of § 3565(a) casts doubt upon the Government’s interpretation. Yet even the Court recognizes that the legislative history is, at best, inconclusive. See ante, at 49 (“None of the legislators’ expressions . . . focuses on ‘the precise meaning of the provision at issue in this case’ ”) (quoting Brief for United States 24, and n. 4); see also ante, at 51-53, and n. 11. Where the language of a statute is clear, that language, rather than “isolated excerpts from the legislative history,” should be followed. Patterson v. Shumate, 504 U. S. 753, 761, and n. 4 (1992).
The Court expresses disbelief that Congress could have intended to authorize punishment for drug-possessing probationers so much more severe than the punishment authorized for the probationer’s original offense. Ante, at 48-49. I think the Court misses two points. First, as the Court itself seems to recognize, the maximum punishment authorized for respondent’s original offense is not the Guidelines range, but the maximum statutory sentence. See 18 U. S. C. §§ 1703(a), 3553(b), 3559(a)(4), and 3581(b)(4). In respondent’s case, the punishment authorized for his original offense is therefore exactly equal to the punishment authorized for his probation violation — five years’ imprisonment. See § 1703(a). Second, Congress provided for equally harsh revocation sentences in the subsections preceding and following § 3565(a). By allowing sentencing courts to impose “any other sentence that was available ... at the time of the initial sentencing,” §§ 3565(a)(2) and (b), Congress authorized these courts to impose the maximum statutory sentence upon revocation of probation. Thus, if respondent’s probation had been revoked pursuant to §§ 3565(a)(2) or (b), he would have faced the same maximum revocation sentence he faces under § 3565(a) — five years’ imprisonment.
The Government suggests that if “original sentence” does not refer to the sentence of probation imposed, then it might just as readily refer to the statutory sentence. The Court rejects this suggestion because imposing the maximum statutory sentence would require an upward departure from the Guidelines range, and probation “is a most unlikely prospect” in any case involving an upward departure. Ante, at 56, n. 14. Thus, according to the Court, it “makes scant sense” to assume that “original sentence” is the statutory maximum sentence. Ibid. By the same reasoning, however, it makes little sense to assume that the maximum Guidelines sentence is the “original sentence,” as probation is an “unlikely prospect” in any case where a defendant would otherwise receive the maximum available sentence under the Guidelines. Indeed, if the plausibility of the potential sentence is the Court’s guide, one would think the Court would choose the bottom of the Guidelines range as its benchmark.