with whom The Chief Justice, Justice White, and Justice Scalia join, dissenting.
Newark Morning Ledger seeks a depreciation1 deduction under 26 U. S. C. § 167(a) for an intangible asset it variously refers to as “paid subscribers,” see Brief for Petitioner 4-5, or “subscriber relationships,” see Tr. of Oral Arg. 3. The Court of Appeals rejected Ledger’s claim on the authority of a Treasury regulation providing (a) that an intangible asset may be depreciated only if it has a limited useful life “the length of which can be estimated with reasonable accuracy,” and (b) that “[n]o deduction for depreciation is allowable with respect to goodwill.” 26 CFR § 1.167(a)-3 (1992); see 945 F. 2d 555, 558, 567-569 (CA3 1991). Ledger claims the regulation raises no bar to a deduction, arguing that (1) the asset is not goodwill, because (2) it has a limited useful life actually estimated with reasonable accuracy. Ledger is wrong on both counts. Ledger’s asset is unmistakably a direct measurement of goodwill, and Ledger’s expert testimony failed to show any particular lifespan for the goodwill Ledger acquired.
I
When The Herald Company (now merged with Newark Morning Ledger) bought and liquidated the stock of Booth Newspapers, Inc., it allocated $67.8 million of the stock’s adjusted basis to an asset called “paid subscribers.” Although, as will appear, this label is misdescriptive, it need not confuse anyone about the true nature of the asset, since Ledger has explained clearly how it determined the asset’s value. Ledger got to the $67.8 million figure by predicting the fu*572ture net revenues to be generated by the 460,000 people who subscribed to the eight Booth newspapers as of the date of sale, May 31, 1977. Because these customers had neither paid in advance nor agreed to subscribe for any set term, see Brief for Petitioner 4, n. 5; ante, at 550, n. 4, they were merely at-will subscribers; the value of their expected future custom was capitalized as the asset Ledger seeks to depreciate.
However much Ledger claims this asset to be something different from “goodwill,” the settled meaning of the term is flatly at odds with Ledger’s contention. Since the days of Justice Story, we have understood the concept of “goodwill” to be anchored in the patronage a business receives from “constant or habitual” customers. See, e. g., Metropolitan Bank v. St. Louis Dispatch Co., 149 U. S. 436, 446 (1893); Des Moines Gas Co. v. Des Moines, 238 U. S. 153 (1915); see also Cruttwell v. Lye, 17 Ves. Jr. 335, 346, 34 Eng. Rep. 129, 134 (Ch. 1810) (opinion of Lord Eldon) (goodwill is “nothing more than the probability, that the old customers will resort to the old place”). Although this Court has not had occasion to provide a precise definition of the term as it appears in the depreciation regulation, the Courts of Appeals have consistently held that “goodwill,” in this context, refers to “the expectancy of continued patronage” from existing customers or, alternatively, to the prospect that “the old customers will resort to the old place.” See, e. g., Winn-Dixie Montgomery, Inc. v. United States, 444 F. 2d 677, 681 (CA5 1971); Commissioner v. Seaboard Finance Co., 367 F. 2d 646, 649 (CA9 1966); Boe v. Commissioner, 307 F. 2d 339, 343 (CA9 1962); Dodge Brothers, Inc. v. United States, 118 F. 2d 95, 101 (CA4 1941); see also Golden State Towel & Linen Service, Ltd. v. United States, 179 Ct. Cl. 300, 305-309, 373 F. 2d 938, 941-943 (1967); Karan v. Commissioner, 319 F. 2d 303, 306 (CA7 1963) (goodwill denotes an expectancy that a customer relationship will continue “without contractual compulsion”). Thus, the Government justifiably concludes that *573“goodwill,” as used in its own regulation, refers to the expectation of continued patronage by existing customers. See Brief for United States 16-19.
Under this accepted definition of “goodwill,” there can be no doubt that the asset Ledger calls “paid subscribers” or “subscriber relationships” is simply the goodwill associated with those subscribers. Once this is clear, it becomes equally clear that Ledger should lose, since the intangible asset regulation expressly and categorically bars depreciation of goodwill, and courts have uniformly relied on that regulation’s plain language to conclude that goodwill is nondepreciable as a matter of law. See Houston Chronicle Publishing Co. v. United States, 481 F. 2d 1240, 1247 (CA5 1973) (the proposition that goodwill is nondepreciable as a matter of law “is so well settled that the only question litigated in recent years regarding this area of the law is whether a particular asset is ‘goodwill’ ”), cert. denied, 414 U. S. 1129 (1974); see also Donrey, Inc. v. United States, 809 F. 2d 534, 536 (CA8 1987) (goodwill “is ineligible per se for the depreciation deduction”); Richard S. Miller & Sons, Inc. v. United States, 210 Ct. Cl. 431, 437, 537 F. 2d 446, 450 (1976) (“the presumption that [goodwill] is a nondepreciable capital asset is conclusive”); Boe v. Commissioner, supra, at 343 (“good will is not a depreciable asset”).
I — ( I — I
Ledger tries to slip out of this predicament by two separate steps. It argues first that the Court ought to adopt a new definition of “goodwill” that would not cover any expectation of future custom with a lifespan subject to definite advance estimate; then it claims that the asset here falls outside the new definition because Ledger’s expert has predicted the length of the asset’s wasting life with reasonable accuracy. See Brief for Petitioner 12-13. The Court makes a serious mistake in taking the first step; Ledger should lose *574in any event, however, since its expert has failed to furnish the basis for taking the second.
A
Ledger would have us scrap the accepted and substantive definition of “goodwill” as an expectation of continued patronage, in favor of a concept of goodwill as a residual asset of ineffable quality, whose existence and value would be represented by any portion of a business's purchase price not attributable to identifiable assets with determinate lives. Goodwill would shrink to an accounting leftover. See id., at 19, 29-30 (relying on accounting standards).
In accommodating Ledger on this point, see ante, at 565, n. 13, the Court abandons the settled construction of a regulation more than 65 years old,2 see T. D. 4055, VI-2 Cum. *575Bull. 63 (1927), and repudiates the equally settled interpretation of the corresponding section of the tax code itself. We are, after all, dealing with a statute reenacted without substantial change not less than six times since 1919, see Revenue Act of 1918, § 234(a)(7), 40 Stat. 1078 (1919); Revenue Act of 1932, §23(k), 47 Stat. 181; Revenue Act of 1934, §23(0, 48 Stat. 689; Revenue Act of 1936, § 23(0, 49 Stat. 1660; Revenue Act of 1938, §23(0, 52 Stat. 462; Internal Revenue Code of 1939, § 23(0, 53 Stat. 14; Internal Revenue Code of 1954, § 167(a), 68A Stat. 51, and we may presume that Congress has accepted the understanding set out in the cognate intangible asset regulation and in the judicial decisions that have clarified that regulation’s terms.3 Lorillard, v. Pons, 434 U. S. 575, 580 (1978); United States v. Correll, 389 U. S. 299, 305-306 (1967); Helvering v. Winmill, 305 U. S. 79, 83 (1938). The consequences, therefore, of acceding to Ledger’s argument are at once a rejection of statutory interpretation settled by Congress itself through reenactment of the tax code and a further invasion of the political domain to rewrite a Treasury regulation.4 See Correll, supra, at 307 (this Court *576will defer to a tax regulation so long as it “implements the congressional mandate in some reasonable manner”); National Muffler Dealers Assn., Inc. v. United States, 440 U. S. 472, 477 (1979) (listing historical considerations that may give a regulation “particular force”); see also Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 843-845 (1984) (reasonable agency interpretations of statutory provisions will be upheld).
I cannot deny, however, that the regulation would suffer real internal tension between its specific, categorical treatment of goodwill and its general analytical test (turning on the existence of a limited life of ascertainable duration), if modern accounting techniques were to develop a subtlety sufficient to make an accurate estimate of goodwill’s useful life. Fortunately or not, however, the record in this case raises no such tension.
B
Even under Ledger’s revision of the regulation, a depreciation deduction would depend on showing the Booth newspapers’ goodwill to have a useful life both limited and measurable with some reasonable degree of certainty. The further step needed for victory is thus evidentiary in nature, and *577Ledger’s success or failure is solely a function of the eviden-tiary record. Ledger has failed.
Here, it is helpful to recall one defining characteristic of the only kind of asset Ledger claims to be entitled to depreciate: it must be an asset acquired from Booth Newspapers, Inc., upon the sale of its stock to Ledger’s predecessor, Herald. If the goodwill is to be depreciated at all, in other words, it must be goodwill purchased, not goodwill attributable to anything occurring after the purchase date. It must be an expectation of continued patronage as it existed when the old Booth newspapers changed hands.
Assuming that there is a variety of goodwill that may be separately identified as an asset on the date of sale, some limitation on its useful life may be presumed. Whatever may be the force of habit, or inertia, that is valued as goodwill attributable to events occurring before the date of sale, the influence of those events wanes over time, and so must the habit or inertia by which that influence is made manifest and valued as goodwill. On the outside, the economically inert subscribers will prove to be physically mortal.5
*578What the Government does not concede,6 however, and what Ledger has not proven, is the duration of that date-of-sale influence and consequent goodwill. Ledger, indeed, has not even purported to show that. Instead, its expert has estimated the quite different periods over which subscribers on the date of sale will continue to subscribe to the various papers.7 In the District Court, Ledger offered a single witness for its claim to have estimated the useful life of each newspaper’s “subscriber relationships” with reasonable accuracy. Herald had originally hired that witness, Dr. Gerald Glasser, to predict the average remaining lives of existing subscriptions to the eight Booth newspapers. See App. in *579No. 90-5637 (CA3), p. 1010. Dr. Glasser testified that he first compiled statistics on the length of time existing subscribers had received each newspaper, by directing a survey that asked a selection of those subscribers one central question: “For how long has the [newspaper] been delivered to your present address?” Id., at 157, 166, 182-183, 1012. He then made a crucial assumption, that the total number of subscriptions to each newspaper would remain stable over time. Id., at 170-172, 187, 194-195. Finally, by subjecting the survey results to techniques of statistical analysis based on this crucial assumption, Dr. Glasser produced a series of figures that, he said, represented the average remaining life of existing subscriptions to each newspaper. Solely on the basis of Dr. Glasser’s testimony, the District Court held that “the remaining useful lives of the paid subscribers of the Booth newspapers as of May 31, 1977, could be estimated with reasonable accuracy.” 734 F. Supp. 176, 181 (NJ 1990).
Dr. Glasser’s assumption is the key not only to the results he derived, but to the irrelevance of those results to the predictable life on the date of sale of the goodwill (or “paid subscribers”) actually purchased from Booth. The key, in turn, to that irrelevance lies not in Dr. Glasser’s explicit statement of his assumption, but in what the assumption itself presupposes. Since the District Court was not concerned with predicting the value that any given Booth newspaper might have in the future (as distinct from predicting the useful life of pre-existing subscriber goodwill), an assumption that the level of a paper’s subscriptions would remain constant was useful only insofar as it had a bearing on predicting the behavior of the old subscribers. For this purpose, assuming a constant subscription level was a way of supposing that a given newspaper would remain as attractive to subscribers in the future as it had been during the period prior to the newspaper’s sale. The assumption was thus a surrogate for the supposition that the new owners would not rock the boat and would succeed in acting intelligently to keep the paper, *580if not exactly as it had always been, at least as relatively attractive as it had been in relation to its various competitors on the date of sale.
What is significant about this assumption for present purposes is not its doubtful validity,8 but the very fact of its being an assumption about the behavior of the paper’s man*581agement after the date of sale. And since this assumption is the basis for a prediction about the life of subscriptions existing on the date of sale, that prediction is by definition not simply about the duration of subscriber goodwill (or habit or inertia) as it existed on the date the paper changed hands. On the contrary, it is a prediction about the combined effect of presale goodwill and postsale satisfaction with the paper as Ledger presumably continues to produce it. Nowhere in Dr. Glasser’s testimony do we find an opinion that the pre-sale goodwill has a life coextensive with the predicted life of the subscriptions, and nowhere do we find an opinion about the point at which the old goodwill finally peters out as a measurable, and hence valuable, influence on the old subscribers’ behavior. It is not, of course, important for present purposes whether such an opinion would be possible, though I am skeptical that it would be.9 But it is important that no such evidence exists in this case. In place of evidence showing the depreciable lifespan of date-of-sale goodwill with a reasonable degree of accuracy, Ledger has presented evidence of how long an old subscriber will remain one, on the assumption that the subscriber’s prior satisfaction is confirmed, and (for all we know) replaced, with satisfaction resulting from Ledger’s publishing performance over the years following its acquisition of a given newspaper.
This, of course, misses the point entirely. In telling us merely how long a subscriber is likely to subscribe, Ledger tells us nothing about how long date-of-sale subscriber habit or inertia will remain a cause of predicted subscriber faithfulness. Since, however, only the date-of-sale probability of faithfulness could be entitled to depreciation as a pur*582chased asset, Ledger’s expert on his own terms has not even claimed to make the showing of definite duration necessary to depreciate an asset under § 167(a). Indeed, once duration of subscriptions and purchased goodwill are seen to be conceptually different, Ledger’s claim to have satisfied the requirements for depreciating an intangible asset simply vanishes. Ledger’s entire case thus rests on the confusion of subscription duration with goodwill on the date of sale, and only that confusion could suggest that Ledger has shouldered its burden of estimating the lifespan of the asset purchased from Booth. It is not surprising, then, that the Commissioner has stood by her categorical judgment that goodwill is not depreciable, that Congress has not disturbed this judgment,10 and that lower courts have consistently agreed that goodwill is nondepreciable as a matter of law.
b — i > — I
Because the Court of Appeals correctly reversed on the basis that Newark Morning Ledger failed to demonstrate that the asset it sought to depreciate was not goodwill, which *583is nondepreciable as a matter of law, see 945 F. 2d, at 568,1 would affirm the judgment below. From the Court’s holding to the contrary, I respectfully dissent.
Black’s Law Dictionary tells us that intangible assets are amortized, while tangible assets are depreciated. Black’s Law Dictionary 83, 441 (6th ed. 1990); see also Gregorcich, Amortization of Intangibles: A Reassessment of the Tax Treatment of Purchased Goodwill, 28 Tax Law. 251, 253 (1975) (“Amortization is the commonly accepted way of referring to depreciation of intangible property”). The statute and the regulations, however, use only the term depreciation.
The current intangible asset regulation can be traced back to Treasury Regulation 45, issued in 1919, which provided that there could be no deduction “in respect of good will” under the general depreciation provision of the Revenue Act of 1918 because goodwill was an example of an asset that did not have a useful life “definitely limited in duration.” T. D. 2831, 21 Treas. Dec. 214, Art. 163. The Commissioner dropped the reference to goodwill for a few years, in response to attempts by distillers and brewers to depreciate goodwill made obsolete by the adoption of the Eighteenth Amendment. See T. D. 2929,1 Cum. Bull. 133 (1919); see also T. D. 3146,23 Treas. Dec. 402, Art. 163 (1920) (reflecting this change). The first Court of Appeals to address the subject, however, held that goodwill could not be depreciated under the Revenue Act of 1918 because it was not susceptible to exhaustion or wear and tear, as required by the statute. Red Wing Malting Co. v. Willcuts, 15 F. 2d 626 (CA8 1926), cert. denied, 273 U. S. 763 (1927). Shortly after that decision, the Commissioner amended the intangible asset regulation by adding the following prohibition: “No deduction for depreciation, including obsolescence, is allowable in respect of good will.” T. D. 4055, VI-2 Cum. Bull. 63 (1927). It has remained there ever since. See, e. g., Treas. Regs. 77, Art. 203 (Revenue Act of 1932); Treas. Regs. 86, Art. 23(0-3 (Revenue Act of 1934); Treas. Regs. 94, Art. 23(0-3 (Revenue Act of 1936); Treas. Regs. 103, § 19.23(0-3 (Internal Revenue Code of 1939); 26 CFR § 1.167(a)-3 (1961) (Internal Revenue Code of 1954).
Although Red Wing Malting provoked a Circuit split, this Court resolved the conflict a few years later by deciding, in line with the Commis*575sioner’s amended regulation, that a brewery could not deduct for the “exhaustion” or “obsolescence” of goodwill as a result of Prohibition. See Clarke v. Haberle Crystal Springs Brewing Co., 280 U. S. 384 (1930); Renziehausen v. Lucas, 280 U. S. 387 (1930); see also V. Loewers Gambrinus Brewery Co. v. Anderson, 282 U. S. 638 (1931) (distinguishing Haberle Springs and allowing a brewery to claim a depreciation deduction for buildings made obsolete by Prohibition).
Legislative materials indicate that Congress is, in fact, aware of the accepted definition of “goodwill.” See, e. g., H. R. Conf. Rep. No. 100-495, p. 937 (1987) (“Goodwill has been defined as the expectancy of continued patronage, for whatever reason, or as the probability that old customers will resort to the old place”).
The majority discounts these consequences by claiming that the utility of the accepted definition of “goodwill” is limited because “[t]he value of every intangible asset is related, to a greater or lesser degree, to the expectation that customers will continue their patronage.” Ante, at 556. But the regulation does not provide that every intangible asset related to goodwill is nondepreciable; rather, it simply states that goodwill itself is *576nondepreciable. Subject to this prohibition, the law concerning the depreciation of intangible assets related to goodwill has developed on a case-by-case basis, and the Government has accepted some of the distinctions that courts have drawn, including the principle that customer lists sold separately from a going business may be depreciable. See Brief for United States 36, n. 34; Rev. Rul. 74-456,1974-2 Cum. Bull. 65,66 (modifying earlier rulings “to remove any implication that customer and subscription lists, location contracts, insurance expirations, etc., are, as a matter of law, indistinguishable from goodwill”). Such matters are not at issue in this case, however, because the asset that Ledger seeks to depreciate is indistinguishable from goodwill. See 945 P. 2d 555, 568 (CA3 1991) (Newark Morning Ledger did not attempt, in this case, to claim a separate depreciation allowance for the subscriber lists it acquired).
While some courts have viewed goodwill as having an indefinite useful life, others have concluded that although goodwill does waste, its useful life cannot be determined with reasonable accuracy. Compare, e. g., Red Wing Malting, 15 F. 2d, at 633 (goodwill is not depreciable because it “does not suffer wear and tear, does not become obsolescent, [and] is not used up in the operation of the business”); Patterson v. Commissioner, 810 F. 2d 562, 569 (CA6 1987) (goodwill “does not waste”); Houston Chronicle Publishing Co. v. United States, 481 F. 2d 1240, 1248 (CA5 1973) (goodwill is “an ongoing asset that fluctuates but does not necessarily diminish”); Landsberger v. McLaughlin, 26 F. 2d 77, 78 (CA9 1928) (goodwill is not subject to exhaustion, wear or tear), with, e. g., Dodge Brothers, Inc. v. United States, 118 F. 2d 95, 100 (CA4 1941) (goodwill is not depreciable because of “manifest difficulties” inherent in estimating its life span); Illinois Cereal Mills, Inc. v. Commissioner, 46 TCM 1001, 1023 (1983), ¶ 83,469 P-H Memo TC (goodwill is not subject to depreciation “because [its] useful life is not susceptible of reasonable estimation”).
In an effort to insulate the case from review, Ledger asserts a concession by the Government below that the asset Ledger wants to depreciate did have a limited useful life that was estimated with reasonable accuracy. Brief for Petitioner 17, and n. 18. The majority does not go quite so far when it observes that “[p]etitioner’s burden in this case was made significantly lighter by virtue of the Government’s litigation strategy.” Ante, at 567. In any event, the District Court’s description of the Government’s strategy makes it clear that the Government has not conceded this case away:
“The parties have agreed that, if the Court determines that the paid subscribers constitute assets which were separate and apart from goodwill and which can be valued separate and apart from goodwill, and if the Court determines that the paid subscribers had useful lives which can be estimated with reasonable accuracy, then the paid subscribers of the Booth newspapers can be depreciated on a straight-line basis over the ... useful lives [shown in the accompanying chart].” 734 F. Supp. 176, 180 (NJ 1990). Thus, the factual concession by the Government came into play only after the District Court rejected two crucial legal arguments: (1) the “paid subscribers” asset is not an asset separate and distinct from goodwill, and (2) the asset did not have a useful life that could be estimated with reasonable accuracy. I find, for the reasons set out in the text, that the District Court erred in rejecting each argument. I also note that a similar litigating strategy did not prevent the Government from prevailing in Haberle Springs. See 280 U. S., at 386 (“The amount of the deduction to be made is agreed upon if any deduction is to be allowed”).
The estimates vary from paper to paper, but I refer to them in the singular, consistently with Ledger’s claim to a singular “asset.”
No matter how much presale satisfaction subscribers have, it seems intuitively obvious that a high enough level of postsale dissatisfaction with a paper would drive subscribers away, as might other postsale events, such as successful competition and demographic changes. The District Court, relying on Ledger’s own witnesses, noted several of the many possible reasons that lead subscribers to cancel their subscriptions:
“Subscribers are lost because of death, relocation, lack of reader time or interest, changing lifestyles, and other factors that are beyond the control of the newspapers. Also, subscribers are lost due to dissatisfaction with the product or service and for various other reasons, including competition from other media sources, such as radio, television, magazines and other paid-circulation and/or free-distribution newspapers.” 734 F. Supp., at 180.
Ledger’s statistician, in effect, made an assumption regarding Ledger’s ability to manage the innumerable factors that keep current customers coming back for more, as well as its ability to attract new customers as the old ones leave. Such discretionary decisions may turn out to be foolish or wise: if foolish, the subscriber base as of the date of sale could be destroyed rapidly; if wise, it would be maintained. The simple recognition that some papers increase their subscriber base over time, while others lose it (and some actually fold), underscores the arbitrariness of the assumption made by Ledger’s expert witness. In any event, Ledger has provided no evidence to support this assumption.
I do not, of course, suggest that a buyer’s treatment of a depreciable asset does not affect the asset’s actual useful life. A machine’s less durable parts must be replaced; it must be oiled, kept from the weather, given fuel, and so on. But there is an identifiable object that endures through time and does not just disappear from inadequate maintenance. Goodwill, on the other hand, can be destroyed rapidly by everything from the nasty personality of a new proprietor to distaste for his publishing policies. Prediction of goodwill’s endurance must always be fraught with a relatively high degree of chance, for discretionary decisions, rather than just ministerial acts (like oiling the gears), must be taken into account.
Goodwill results from such a mix of influences over time that it seems unlikely that the skein of them all could be untangled to identify the degree to which even present custom results from the goodwill purchased, as distinct from goodwill subsequently cultivated. Ledger has not even attempted such a disentanglement.
The majority claims its approach to be “more faithful to the purposes of the Code,” in allowing taxpayers to make a better match of expenses with revenues. Ante, at 565 (citing INDOPCO, Inc. v. Commissioner, 503 U. S. 79 (1992)). Such policy initiatives are properly left to Congress, which can modify the per se ban on depreciating goodwill at any time. Despite several recent opportunities to do so, Congress has so far refused to alter the tax treatment of goodwill and other intangibles. See, e. g., H. R. 11, 102d Cong., 2d Sess., § 4501 (1992) (as returned from conference, Oct. 5, 1992) (proposing amortization of purchased goodwill and certain other intangible assets over a 14-year period); H. R. 4210, 102d Cong., 2d Sess., §4501 (1992) (as returned from conference, Mar. 20, 1992) (same); H. R. 3040, 102d Cong., 2d Sess., § 302 (1992) (as returned from the Committee on Finance, June 19,1992) (16-year period); H. R. 3035,102d Cong., 1st Sess., § 1 (1991) (as introduced, July 25,1991) (14-year period); see also H. Res. 292,102d Cong., 1st Sess. (1991) (adopted Nov. 26, 1991, 137 Cong. Rec. H11317-H11318) (concerning the effective date of “any legislation enacted with respect to amortization of goodwill”).