with whom The Chief Justice and Justice O’Connor join, dissenting.
Today the Court holds that petitioner Fall River Dyeing & Finishing Corp. violated §§ 8(a)(1) and (a)(5) of the National Labor Relations Act, 29 U. S. C. §§ 158(a)(1), (5) (NLRA), by refusing to bargain with a union that claims to represent its workers. The Court agrees with the National Labor Relations Board (NLRB or Board) that this duty to bargain arose because petitioner is a “successor” to Sterlingwale Corp., a defunct entity that had engaged in a similar line of business. The Court also agrees that the duty to bargain arose when petitioner had brought its first shift into full operation. The theory is that petitioner then had hired a “substantial and representative complement” of its work force. In my view, the Court has misconstrued the successorship doctrine and misapplied the substantial complement test. Accordingly, I dissent.1
*55i — i
!>
Although the Court describes the background of this case in great detail, it gives insufficient consideration to a number of critical facts. On February 12,1982, a financially troubled Sterlingwale ceased operations and indefinitely laid off its production workers, retaining only a skeleton crew to ship out the remaining orders and liquidate the inventory. The collective-bargaining agreement (CBA) between the union and Sterlingwale was allowed to expire in April, and the company ceased paying the workers’ life and health insurance premiums. Attempts to, obtain new financing to keep the business afloat were unsuccessful. Sterlingwale commenced its liquidation by making an assignment for the benefit of creditors, and then hired a professional liquidator to sell the remaining assets at a public auction. By mid to late summer of 1982, all business activity had ceased, and the company permanently closed its doors.
Petitioner Fall River Dyeing & Finishing Corp. was incorporated at the end of August 1982. It bought most of Sterlingwale’s machinery, furniture, and fixtures. It also bought a portion of the Sterlingwale inventory at the public auction.2 Three weeks later it began recruiting new employees by placing ads in the local newspaper. Petitioner hired some former Sterlingwale workers, although by no means all or even a large percentage of those who had been laid off.3 When making its hiring decisions, petitioner took *56into account the applicant’s experience with either Sterling-wale or other finishing plants, App. 223; although the former Sterlingwale supervisors who had been hired were consulted as to the former Sterlingwale workers who applied, there is no finding that these workers as a group received a hiring preference.- Once the new company began operations in November 1982, it performed commission finishing work exclusively, rather than the converting finishing that had accounted for 60%-70% of Sterlingwale’s business.4
B
Of course, a decision by the NLRB that one company is a successor of another is entitled to deference, and its conclusions will be upheld if they are based on substantial record evidence. See Golden State Bottling Co. v. NLRB, 414 U. S. 168, 181 (1973). The critical question in determining successorship is whether there is “substantial continuity” between the two businesses. Aircraft Magnesium, Division ofGrico Corp., 265 N. L. R. B. 1344, 1345 (1982), enf’d, 730 F. 2d 767 (CA9 1984). See also NLRB v. Burns International Security Services, Inc., 406 U. S. 272, 279-281 (1972). Here the Board concluded that there was sufficient *57continuity between petitioner and Sterlingwale, primarily because the workers did the same finishing work on the same equipment for petitioner as they had for their former employer. See 272 N. L. R. B. 839, 840 (1984) (decision of Administrative Law Judge (ALJ)). In reaching this conclusion, however, the Board, and now the Court, give virtually no weight to the evidence of discontinuity, that I think is overwhelming.
In this case the undisputed evidence shows that petitioner is a completely separate entity from Sterlingwale. There was a clear break between the time Sterlingwale ceased normal business operations in February 1982 and when petitioner came into existence at the end of August.5 In addition, it is apparent that there was no direct contractual or other business relationship between petitioner and Sterling-wale. See App. 205. Although petitioner bought some of Sterlingwale’s inventory, it did so by outbidding several other buyers on the open market. Also, the purchases at the public sale involved only tangible assets. Petitioner did not buy Sterlingwale’s trade name or goodwill, nor did it assume any of its liabilities. And while over half of petitioner’s business (measured in dollars) came from former Sterlingwale customers, apparently this was due to the new company’s skill in marketing its services. There was no sale or transfer of customer lists, and given the 9-month interval between the time that Sterlingwale ended production and petitioner commenced its operations in November, the natural conclusion is that the new business attracted customers through its own efforts. No other explanation was offered. Cf. Lincoln Private Police, Inc., 189 N. L. R. B. 717, 719 (1971) (finding it relevant to the successorship question that, while the new *58business acquired many of the former company’s clients, “it did so by means of independent solicitation”). Any one of these facts standing alone may be insufficient to defeat a finding of successorship, but together they persuasively demonstrate that the Board’s finding of “substantial continuity” was incorrect.6
The Court nevertheless is unpersuaded. It views these distinctions as not directly affecting the employees’ expectations about their job status or the status of the union as their representative, even though the CBA with the defunct corporation had long since expired. See Golden State Bottling Co. v. NLRB, supra, at 184 (emphasizing the importance of the workers’ perception that their job situation continues “essentially unaltered”). Yet even from the employees’ perspective, there was little objective evidence that the jobs with petitioner were simply a continuation of those at Sterlingwale. When all of the production employees were laid off indefinitely in February 1982, there could have been little hope — and certainly no reasonable expectation — that Sterlingwale would ever reopen. Nor was it reasonable for the employees to expect that Sterlingwale’s failed textile operations would be resumed by a corporation not then in existence. The CBA had expired in April with no serious effort to renegotiate it, and with several of the employees’ benefits left unpaid. The possibility of further employment with Sterlingwale then disappeared entirely in August 1982 when *59the company liquidated its remaining assets. Cf. Textile Workers v. Darlington Manufacturing Co., 380 U. S. 263, 274 (1965) (the “closing of an entire business . . . ends the employer-employee relationship”). After petitioner was organized, it advertised for workers in the newspaper, a move that hardly could have suggested to the old workers that they would be reinstated to their former positions. The sum of these facts inevitably would have had a negative “effect on the employees’ expectations of rehire.” See Aircraft Magnesium, 265 N. L. R. B., at 1346. See also Radiant Fashions, Inc., 202 N. L. R. B. 938, 940 (1973). The former employees engaged by petitioner found that the new plant was smaller, and that there would be fewer workers, fewer shifts, and more hours per shift than at their prior job. Moreover, as petitioner did not acquire Sterlingwale’s personnel records, the benefits of having a favorable work record presumably were lost to these employees.
In deferring to the NLRB’s decision, the Court today extends the successorship doctrine in a manner that could not have been anticipated by either the employer or the employees. I would hold that the successorship doctrine has no application when the break in continuity between enterprises is as complete and extensive as it was here.
( — I I — I
Even if the evidence of genuine continuity were substantial, I could not agree with the Court’s decision. As we have noted in the past, if the presumption of majority support for a union is to survive a change in ownership, it must be shown that there is both a continuity of conditions and a continuity of work force. Howard Johnson Co. v. Hotel Employees, 417 U. S. 249, 263 (1974). This-means that unless a majority of the new company’s workers had been employed by the former company, there is no justification for assuming that the new employees wish to be represented by the former union, or by any union at all. See Spruce Up Corp., 209 *60N. L. R. B. 194, 196 (1974), enf’d, 529 F. 2d 516 (CA4 1975); 209 N. L. R. B., at 200 (member Kennedy, concurring in part and dissenting in part); Saks & Co. v. NLRB, 634 F. 2d 681, 685-686 (CA2 1980). Indeed, the rule hardly could be otherwise. It would be contrary to the basic principles of the NLRA simply to presume in these cases that a majority of workers supports a union when more than half of them have never been members, and when there has been no election.
The Court acknowledges that when petitioner completed the employment of its anticipated work force in April 1983, less than 50% of its employees formerly had worked for Sterlingwale. It nevertheless finds that the new company violated its duty to bargain, because at an earlier date chosen by the Board, a majority of the work force formerly had worked for Sterlingwale. The NLRB concluded that even though petitioner was still in the process of hiring employees, by the middle of January it had hired a “substantial and representative complement,” when its first shift was adequately staffed and most job categories had been filled.
In my view, the Board’s decision to measure the composition of petitioner’s work force in mid-January is unsupportable. The substantial and representative complement test can serve a useful role when the hiring process is sporadic, or the future expansion of the work force is speculative. But as the Court recognized in NLRB v. Burns International Security Services, Inc., in some cases “it may not be clear until the successor employer has hired his full complement of employees that he has a duty to bargain with a union, since it will not be evident until then that the bargaining representative represents a majority of the employees in the unit.” 406 U. S., at 295. Indeed, where it is feasible to wait and examine the full complement — as it was here — it clearly is fairer to both employer and employees to do so. The substantial complement test provides no more than an estimate of the percentage of employees from the old company that eventually will be part of *61the new business, and thus often will be an imperfect measure of continuing union support. The risks of relying on such an estimate are obvious. If the “substantial complement” examined by the Board at a particular time contains a disproportionate number of workers from the old company, the result either might be that the full work force is deprived of union representation that a majority favors, or is required to accept representation that a majority does not want. Accordingly, unless the delay or uncertainty of future expansion would frustrate the employees’ legitimate interest in early representation — a situation not shown to exist here — there is every reason to wait until the full anticipated work force has been employed.
In this case the date chosen by the NLRB for measuring the substantial complement standard is unsupportable, and the Court’s affirmance of this choice, curious. In prior decisions, courts and the Board have looked not only to the number of workers hired and positions filled on a particular date, but also to “the time expected to elapse before a substantially larger complement would be at work ... as well as the relative certainty of the employer’s expected expansion.” Premium Foods, Inc. v. NLRB, 709 F. 2d 623, 628 (CA9 1983). See also St. John of God Hospital, Inc., 260 N. L. R. B. 905 (1982). Here the anticipated expansion was both imminent and reasonably definite. The record shows that in January petitioner both expected to, and in fact subsequently did, hire a significant number of new employees to staff its second shift. Although the Court finds that the growth of the work force was “contingent” on business conditions, neither the ALJ nor the NLRB made such a finding.7 In fact, they both noted that by January 15, the second shift already had begun limited operations. See 272 N. L. R. B., at 839, n. 1 (“In *62mid-January [petitioner] had one shift in full operation and had started a second shift”); id., at 840. In fact, less than three months after the duty to bargain allegedly arose, petitioner had nearly doubled the size of its mid-January work force by hiring the remaining 50-odd workers it needed to reach full production. This expansion was not unexpected; instead, it closely tracked petitioner’s original forecast for growth during its first few months in business.8 Thus there was no reasonable basis for selecting mid-January as the time that petitioner should have known that it should commence bargaining.9
As the Court notes, the substantial complement rule reflects the need to balance “the objective of insuring maximum employee participation in the selection of a bargaining agent against the goal of permitting employees to be represented as quickly as possible.” Ante, at 48 (citations and internal *63quotation marks omitted). The decision today “balances” these interests by overprotecting the latter and ignoring the former. In an effort to ensure that some employees will not be deprived of representation for even a short time, the Court requires petitioner to recognize a union that has never been elected or accepted by a majority of its workers. For the reasons stated, I think that the Court’s decision is unfair both to petitioner, who hardly could have anticipated the date chosen by the Board, and to most of petitioner’s employees, who were denied the opportunity to choose their union. I dissent.
As a preliminary matter, the Court holds that if one company is a successor to another, it has an obligation to bargain with the prior company’s union even though that union had not been certified recently by the workers. Ante, at 41. As the Court notes, the finding of successorship in NLRB v. Burns International Security Services, Inc., 406 U. S. 272 (1972), was based partly on the fact that the union had been certified almost immediately before the employees were hired by the new company. Id., at 278. Although the Court concludes that the successorship doctrine is not limited to such cases, it certainly would be reasonable to assume that the more remote the certification, the weaker the presumption should be that the union retains majority support. In any event, I do not reach the *55issue because I think it is clear that petitioner is not a successor to Sterlingwale.
At the auction, that apparently took place in October 1982, petitioner bought $13,000 worth of inventory out of the $30,000 worth that was sold. App, 200-201.
More than 150 workers were laid off in February 1982; by mid-January 1983, petitioner had hired perhaps 36 of them. See App. 92-93; 775 F. 2d 425, 428 (CA1 1985). The record does not reveal how many of the laid-off Sterlingwale workers applied for positions with the new company, al*56though petitioner’s vice president testified that he received what “seemed like thousands” of applications in response to the ads. App. 198.
The Court finds little significance in this switch from converting to commission work, since the change was thought to have no direct effect on the employer-employee relationship. See ante, at 46, n. 11. This difference alone would not be determinative, but it hardly is irrelevant. The change meant that unlike petitioner, Sterlingwale did not have to maintain a sales force or retail outlet to sell its cloth, nor did it have to allocate capital for purchasing material. These facts are pertinent to the question whether there is substantial continuity between the two enterprises. The Board in the past has recognized the significance of similar considerations that have an indirect impact on the workers. See, e. g., Gladding Corp., 192 N. L. R. B. 200, 202 (1971) (change in suppliers); Radiant Fashions, Inc., 202 N. L. R. B. 938, 940 (1973) (substantial change in identity of customers).
The Court dismisses the effect of this 7-month hiatus, stating that such a break is important only if there are “other indicia of discontinuity.” Ante, at 45 (citing NLRB v. Band-Age, Inc., 534 F. 2d 1, 5 (CA1), cert. denied, 429 U. S. 921 (1976)). Of course, as noted in the text, there are a number of other “indicia of discontinuity” in this case.
The case before us bears a substantial resemblance to Radiant Fashions, Inc., supra. In that case, the alleged successor was engaged in a business similar to that of its predecessor, at the same location, with the same equipment, the same supervisory personnel, and a reduced but similar work force. The Board nevertheless ruled that the company was not a successor. It based its conclusion on four factors: (i) there was a “lengthy” hiatus of 27a to 3 months between the time the first company shut down and the second company began production; (ii) the second company bought only the assets of the first business, rather than an ongoing enterprise; (iii) the second company served a different market; and (iv) there was no transfer of customers as a result of the sale. 202 N. L. R. B., at 940-941.
The evidence shows that in the textile industry, two shifts are necessary for proper finishing work. See 775 F. 2d 425, 428 (CA1 1985). See also App. 227. Thus, it was clear in mid-January that petitioner would need more employees in the immediate future.
Petitioner’s vice president of operations testified:
“We planned to have a full one shift operation of 56 to 60 employees. And after we reached that goal, and then we’d see how business would be, and then we’d had [sic] planned that by the end of March, April, we should be in a full two shift operation and up to our expected production.” App. 208.
There is no evidence that business conditions during this period were such that the company considered changing its hiring goal.
The NLRB’s reliance on the substantial complement standard is particularly puzzling on these facts, since the evidence shows that the “substantial” complement examined by the Board was not truly “representative” of the work force. When the unfair labor practice hearing was held on May 2, 1983, petitioner already had hired a full complement of workers. At that point the company employed 106-109 workers, less than half of whom were former Sterlingwale employees. Rather than rely on this accurate measure of the composition of the work force, the ALJ looked back to the middle of January, and concluded that petitioner should have acted differently because it appeared at the time that most of the workers who eventually would be represented by the union would be ex-Sterlingwale employees. In other words, the Board ruled that petitioner violated the NLRA because it failed to make the same estimate in January that the ALJ made in May — an estimate that already had proved to be erroneous at the time that the ALJ made it.