dissenting.
I cannot deny that the Court’s opinion persuasively defends a possible interpretation of 26 U. S. C. § 1563(a)(2). In my view, however, the Court has totally failed to establish that the Commissioner’s interpretation is incorrect. Because I believe that the only certainty about the language and history of § 1563(a)(2) is that both are ambiguous, I would defer to the Commissioner’s judgment.
The Court begins by declaring that the statutory language, “while not completely unambiguous, is in closer harmony with the taxpayer’s interpretation than with the Commissioner’s Regulation” because the term “ ‘brother-sister controlled group’ — connotes a close horizontal relationship between two or more corporations.” Ante, at 25 (emphasis in original). In taking this approach, however, the Court simply assumes its conclusion. The 50-percent test of Part (B) already ensures a horizontal relationship between the corporations that constitute the controlled group; nothing in the language of the statute suggests that Part (A) was designed directly to serve the same purpose. At most, § 1563(a)(2) can be read to require that the same set of five or fewer persons must satisfy the 50- and 80-percent tests; the statute is entirely silent as to whether each member of the set must own stock in each corporation. And, unlike the Court, I have difficulty inferring this conclusion from the term “brother-sister controlled group,” a phrase that appears only in the heading of the subsection and that is hardly a household term with an intuitively obvious meaning.
Similar problems attend the Court’s analysis of the statute’s structure. In the Court’s view, the fact that the controlling group of shareholders is defined as “5 or fewer” for both the 50- and 80-percent tests “suggests that precisely the *36same shareholders must satisfy both the 80-percent and 50-percent requirements.” Ante, at 25. Even if this were true, however, it would not mean that each member of the set of five or fewer shareholders must own stock in each corporation; it suggests only that the total number of shareholders considered in relation to both tests may not exceed five. In any event, the common-ownership requirement — which takes “into account the stock ownership of each such person only to the extent such stock ownership is identical with respect to each such corporation,” § 1563(a)(2)(B) — is embedded in Part (B), and the simpler and normal reading of the statute therefore would apply the common-ownership restriction only to Part (B)’s 50-percent test.1 It is the Court’s reading, then, that seemingly runs counter to the structure of the statute, for under its approach the 80-percent test would “tend to overlap or swallow the 50% identical ownership requirement.” Allen Oil Co. v. Commissioner, 614 F. 2d 336, 339 (CA2 1980).
The confusing nature of the statutory text leads the Court to rely principally on § 1563(a)(2)’s legislative history, which it cheerfully reads as “resolvfing] any ambiguity in the statutory language.” Ante, at 26. It seems to me that this conclusion is substantially overstated. It is undoubtedly true, as the Court observes, that § 1563(a)(2) was aimed at curbing the abuses of multiple incorporation. But this is beside the *37point, for — as the Court notes — the 50-percent test of Part (B) itself serves to “measur[e] . . . the overlap between two corporations.” Ante, at 27. The Court’s further conclusion “that Congress intended that the 80-percent requirement be the primary requirement for defining the interrelationship between two or more corporations,” ibid, (emphasis in original), is entirely without support in the legislative history.2 Certainly, such a view appears nowhere in the congressional Reports. These simply echo the statutory definition, declaring that a controlled group includes “two or more corporations which are owned 80 percent or more ... by five or fewer persons . . . provided that these five or fewer persons own more than 50 percent of each corporation when the stock of each person is considered only to the extent it is owned identically with respect to each corporation.” H. R. Rep. No. 91-413, pt. 1, p. 99 (1969). Accord, S. Rep. No. 91-552, p. 135 (1969). Again, however, the legislative documents prove only that the same set must satisfy the 80- and 50-per-*38cent tests; they cannot easily be read to require that each member of the set own stock in every corporation.
Ironically, then, the Court at bottom is forced to rely on the rationale advanced by the Treasury Department when it proposed the legislation eventually adopted as § 1563(a)(2). The Court’s analysis of this proposal, which it explores in some detail, ante,- at 28-30, is certainly credible. But even this legislative material contains an essential ambiguity.3 Neither the “General Explanation” nor the “Technical” one addresses whether the 80-percent test requires common ownership, or whether a person excluded from the 50-percent calculation because he owns no stock in one of the controlled corporations may nevertheless be included in the 80-percent test, so long as the total number of relevant shareholders does not exceed five. For example, while the Treasury Department suggested that “the same five or fewer persons [must] own at least 80 percent of the voting stock or value of shares of each corporation” to satisfy Part (A), and that “these five or fewer individuals” must satisfy the 50-percent test of Part (B), Hearings Before the House Committee on Ways and Means on the Subject of Tax Reform, 91st Cong., 1st Sess., pt. 14, p. 5168 (1969) (emphasis in original), the Department’s explanation — despite the Court’s suggestion to the contrary — need not be read as requiring that each of the five own stock in every controlled corporation. To the contrary, the Technical Explanation declares that the 80-percent test “is satisfied if the group of five or fewer persons as a whole owns at least 80 percent of the voting stock or value of shares of each corporation, regardless of the size of the indi*39vidual holdings of each person.” Id., at 5169 (emphasis added). This obviously suggests that the crucial inquiry is whether a given set of five satisfies both tests, not whether each individual owns stock in each corporation.
Certainly, I do not suggest that the Commissioner’s interpretation is compelled by the legislative materials. But the Court, by putting so much effort into reading between the lines, has lost sight of the fact that certain statutory ambiguities cannot be neatly and finally resolved. Here, the Commissioner’s interpretation is not “unreasonable or meaningless,” for “it insures that the stock is closely held.” Allen Oil Co. v. Commissioner, 614 F. 2d, at 340. In such a situation, “[tjhe choice among reasonable interpretations is for the Commissioner, not the courts.” National Muffler Dealers Assn., Inc. v. United States, 440 U. S. 472, 488 (1979). See United States v. Correll, 389 U. S. 299, 307 (1967). For that reason, I respectfully dissent.
The Court concludes that the phrase “each such person” in Part (B) refers back to the “5 or fewer persons,” which precedes Part (A), “strengthening the suggestion that there is one fixed, indivisible group of shareholders whose holdings are to be considered throughout application of both the 80-percent requirement in Part (A) and the 50-percent requirement in Part (B).” Ante, at 26, n. 8. But this language proves only that the total number of shareholders considered may not exceed five; it need not be read to require that each 80-percent shareholder own stock in each corporation. Indeed, the presence of an explicit common-ownership requirement in Part (B), along with the absence of analogous language in Part (A), suggests that Congress did not intend to write such a requirement into the 80-per-cent test.
The Court apparently derives this conclusion from the nature of the pre-1969 statutory scheme, under which corporations were considered to be part of a controlled group only if the same person owned 80 percent of the stock in each controlled corporation. Ante, at 28. In the Court’s view, § 1563(a)(2) simply expanded the ownership group to five, retaining the 80-percent requirement as the primary test for interrelatedness. The problem with this approach is that it is entirely speculative. Congress nowhere stated that it had any such intention with regard to the 80-percent test. And the Treasury Department, when it proposed § 1563(a)(2), simply stated the obvious: it declared that the new statute “expand[ed] present law” by considering the ownership interests of five individuals, while adding a 50-pereent test “to insure” that controlled corporations operate as one economic entity. Hearings Before the House Committee on Ways and Means on the Subject of Tax Reform, 91st Cong., 1st Sess., pt. 14, p. 5394 (1969). Certainly, the Court can credibly read its conclusion into this history. But the legislative materials are not inconsistent with the Commissioner’s contrary view that the newly devised 50-percent test was to serve as the primary indicium of interrelatedness. Because of the absence of any explicit statement on the question in the legislative history, I find the Court’s certainty somewhat surprising.
Indeed, throughout the course of litigation over § 1563(a)(2), both the Commissioner and the various taxpayers involved have drawn support from precisely the same portions of the Treasury Department proposals. Compare Fairfax Auto Parts of Northern Virginia, Inc. v. Commissioner, 65 T. C. 798, 803-804 (1976), rev’d, 548 F. 2d 501 (CA4), cert. denied, 434 U. S. 904 (1977), with 65 T. C., at 809-810 (dissenting opinion). See also Allen Oil Co. v. Commissioner, 614 F. 2d 336, 340, n. 4 (CA2 1980).