I agree with the majority. I write separately to state another reason the trial court correctly granted summary judgment against plaintiff John Guz on the age discrimination claim: Even after “extensive discovery” (maj. opn., ante, at p. 327), Guz has produced no credible evidence that defendants Bechtel National, Inc., and Bechtel Corporation (collectively Bechtel) discharged him because of his age. Bechtel, the moving party on summary judgment, has met its burden of showing that Guz cannot state a prima facie age discrimination case. Accordingly, Bechtel had no duty even to rebut Guz’s age discrimination claim, although I agree that it also did so.
*372I. The Legal Standard for Summary Judgment
To prevail at trial, indeed, to avoid a nonsuit, the plaintiff bears the burden of establishing a prima facie case of discrimination. (See generally McDonnell Douglas Corp. v. Green (1973) 411 U.S. 792 [93 S.Ct. 1817, 36 L.Ed.2d 668]; maj. opn., ante, at pp. 354-355; Caldwell v. Paramount Unified School Dist. (1995) 41 Cal.App.4th 189, 203-204 [48 Cal.Rptr.2d 448].) I agree with the majority regarding what this prima facie burden is. It is not onerous, but the plaintiff must show that the employer’s actions, if unexplained, support an inference that they were more likely than not based on a prohibited discriminatory criterion. (Maj. opn., ante, at p. 355.) Specifically, the plaintiff must show some “circumstance [that] suggests discriminatory motive.” (Ibid.; see also O’Connor v. Consolidated Coin Caterers Corp. (1996) 517 U.S. 308, 312 [116 S.Ct. 1307, 1310, 134 L.Ed.2d 433].)1
Some uncertainty currently exists regarding the way this rule applies to an employer’s motion for summary judgment in a discrimination action. (Maj. opn., ante, at pp. 356-357.) California’s traditional rule was that to prevail on summary judgment, a “defendant must conclusively negate a necessary element of the plaintiff’s case, and demonstrate that under no hypothesis is there a material issue of fact that requires the process of a trial.” (Molko v. Holy Spirit Assn. (1988) 46 Cal.3d 1092, 1107 [252 Cal.Rptr. 122, 762 P.2d 46].) In 1992 and 1993, however, the Legislature amended Code of Civil Procedure section 437c, the statute concerning summary judgment. (See generally Union Bank v. Superior Court (1995) 31 Cal.App.4th 573, 581-584 [37 Cal.Rptr.2d 653].) Today, as relevant, Code of Civil Procedure section 437c, subdivision (o)(2), provides that a defendant has met its burden on summary judgment “of showing that a cause of action has no merit if that party has shown that one or more elements of the cause of action . . . cannot be established, or that there is a complete defense to that cause of action. Once the defendant . . . has met that burden, the burden shifts to the plaintiff... to show that a triable issue of one or more material facts exists as to that cause of action or a defense thereto. The plaintiff. . . may not rely upon the mere allegations or denials of its pleadings to show that a triable issue of material fact exists but, instead, shall set forth the specific facts showing that a triable issue of material fact exists as to that cause of action or a defense thereto.”
*373This court has not yet considered the effect of these amendments in a discrimination case. The Courts of Appeal have, however, considered this question in detail, and not always consistently. (See, e.g., the exhaustive discussion in Scheiding v. Dinwiddle Construction Co. (1999) 69 Cal.App.4th 64, 69-83 [81 Cal.Rptr.2d 360] (Scheiding).) Some courts have held that to prevail on summary judgment, the defendant need merely point to the plaintiff’s lack of evidence establishing a prima facie case. (Hersant v. Department of Social Services (1997) 57 Cal.App.4th 997, 1002 [67 Cal.Rptr.2d 483] [“The burden-shifting system requires the employee first establish a prima facie case of age discrimination”]; Horn v. Cushman & Wakefield Western, Inc. (1999) 72 Cal.App.4th 798, 806 [85 Cal.Rptr.2d 459]; Caldwell v. Paramount Unified School Dist., supra, 41 Cal.App.4th at p. 203 [“Thus, the burdens of proof for purposes of a defendant’s motion for summary judgment are precisely the same as those mandated by McDonnell Douglas”).) Others have required the defendant to prove the “plaintiff’s inability to prove its own case . . . .” (Certain Underwriters at Lloyd’s of London v. Superior Court (1997) 56 Cal.App.4th 952, 959 [65 Cal.Rptr.2d 821], italics omitted.) Others have suggested that because the moving party must negate the plaintiff’s right to prevail on a particular issue, the burden is reversed on summary judgment. (Sada v. Robert F. Kennedy Medical Center (1997) 56 Cal.App.4th 138, 150 [65 Cal.Rptr.2d 112]; Addy v. Bliss & Glennon (1996) 44 Cal.App.4th 205, 216 [51 Cal.Rptr.2d 642].)
I believe the Court of Appeal cases can generally be reconciled. The recent decisions recognize that Code of Civil Procedure section 437c, subdivision (o)(2), significantly changed California summary judgment law. To prevail on summary judgment, the defendant no longer must conclusively negate the plaintiff’s case. Given the difficulty of proving a negative, such a test is often impossibly high. However, the statute also places an initial burden on the defendant in order to prevail on summary judgment. As explained in Scheiding, the differences in the cases can largely be described as differing degrees of caution rather than outright disagreement. (Scheiding, supra, 69 Cal.App.4th at pp. 82-83.) That case quoted with approval (ibid.) most of the following discussion in Hagen v. Hickenbottom (1995) 41 Cal.App.4th 168 [48 Cal.Rptr.2d 197]: “We cannot agree with those who may be understood to suggest that a moving defendant may shift the burden simply by suggesting the possibility that the plaintiff cannot prove its case. It is clear to us, from the requirement . . . that a defendant have ‘shown that one or more elements of the cause of action . . . cannot be established’ (Code Civ. Proc., § 437c, former subd. (n)(2) [now subd. (g)(2)]; . . .), that a defendant must make an affirmative showing in support of his or her motion. Such a showing connotes something significantly more than simply ‘pointing out to the . . . court’ that ‘there is an absence of evidence’: before *374the burden of producing even a prima facie case should be shifted to the plaintiff in advance of trial, a defendant who cannot negate an element of the plaintiff’s case should be required to produce direct or circumstantial evidence that the plaintiff not only does not have but cannot reasonably expect to obtain a prima facie case. But where such a showing can be made we consider it both fair to the defendant and consistent with efficient administration of justice that the plaintiff be called upon, on risk of summary judgment, to make a prima facie case.” (Id. at p. 186.)
I think this discussion aptly summarizes the law. To prevail on summary judgment in a discrimination case, the defendant must show that the plaintiff both has not established and cannot reasonably expect to establish a prima facie case. A defendant can meet the former burden merely by showing the absence of evidence of discrimination. But that is not enough. The defendant must also show, by direct or circumstantial evidence, that the plaintiff cannot reasonably expect to obtain a prima facie case. This latter showing, however, is not impossibly difficult. If a plaintiff has had the full opportunity to obtain discovery and to present all available evidence in support of a discrimination claim, and still has failed to establish a prima facie case, the trial court may reasonably infer that the plaintiff cannot do so. If the plaintiff cannot present a prima facie case, a nonsuit at trial would be inevitable. (Code Civ. Proc., § 581c; Caldwell v. Paramount Unified School Dist, supra, 41 Cal.App.4th at pp. 203-204.) In that case, the trial court should grant summary judgment and avoid a useless trial.
II. The Legal Standard Applied to This Case
Bechtel has shown that Guz had a full opportunity to discover and present all available evidence, and that he nonetheless has not stated a prima facie case of age discrimination. This showing meets Bechtel’s burden of establishing that Guz cannot state a prima facie case, thus entitling it to summary judgment on this cause of action. Despite extensive discovery, Guz can point to no comments by anyone during his entire lengthy history with Bechtel suggesting age played a role in employment decisions in general, or in his case in particular, no meaningful statistical evidence, no evidence even of a possible financial or other motive for Bechtel to get rid of its older workers. He cites no evidence suggesting that age was a significant factor in Bechtel’s layoff decision.2
It is true that Guz was a member of the protected class, but that fact alone proves nothing. In a reduction in force, many qualified, productive workers, *375both within and outside a protected class, lose their positions. Older workers may be laid off just like younger ones. The laws against age discrimination do not “ ‘required that younger employees be fired so that employees in the protected age group can be hired.’ ” (Earley v. Champion Intern. Corp. (11th Cir. 1990) 907 F.2d 1077, 1083; see also Vaughan v. MetraHealth Companies, Inc. (4th Cir. 1998) 145 F.3d 197, 204 [age discrimination laws are not “something akin to a strict seniority protection system”]; Jameson v. Arrow Co. (11th Cir. 1996) 75 F.3d 1528, 1532-1533.) Accordingly, “the decision to discharge a qualified, older employee is not inherently suspicious. ... In a [reduction in force], qualified employees are going to be discharged.” (Brocklehurst v. PPG Industries, Inc. (6th Cir. 1997) 123 F.3d 890, 896.)
Like the concurring and dissenting opinion, Guz relies largely on two circumstances to support the age discrimination claim. The first is that, of three positions that employees other than Guz filled while Guz was on holding status, younger employees filled two, and an older employee filled only one. This fact is meaningless. Even aside from the minuscule size of the sampling (see post), Guz does not tell us the average age of persons eligible or considered for these positions. If a majority of those persons were younger than Guz, a majority of those given the positions would likely also be younger. Guz “did not even attempt to place his figures in a relevant context so as to make them meaningful. . . . HO . . . [H]e neglects vital information regarding the pool of applicants and whether, for example, qualified older employees were available or applied for those jobs. . . . [Ejmployee statistics unaccompanied by evidence regarding qualified potential applicants from the relevant labor market . . . lack[] probative value.” (Simpson v. Midland-Ross Corp. (6th Cir. 1987) 823 F.2d 937, 943.) Guz’s related arguments regarding his qualifications relative to those retained does not aid him. “As courts are not free to second-guess an employer’s business judgment, this assertion [that plaintiff was equally or more qualified than the people retained] is insufficient to permit a finding of pretext.” (Branson v. Price River Coal Co. (10th Cir. 1988) 853 F.2d 768, 772.) “Thus, plaintiff’s general dispute concerning his job performance, in the absence of any other evidence of age discrimination, does not provide a sufficient basis for a jury to infer that [the employer] terminated plaintiff on the basis of his age.” (Fallis v. Kerr-McGee Corp. (10th Cir. 1991) 944 F.2d 743, 747.)
The second circumstance Guz cites comes closest to presenting evidence that might suggest age discrimination: Of the six persons in his unit, Bechtel retained the youngest two. This fact also fails to arouse suspicion for several reasons.
First, a group of six is simply too small to be statistically significant. “For [the plaintiff] to show a prima facie case of disparate treatment based solely *376on statistics he must show a 1 “stark” pattern’ of discrimination unexplainable on grounds other than age.” (Palmer v. United States (9th Cir. 1986) 794 F.2d 534, 539; Rose v. Wells Fargo & Co. (9th Cir. 1990) 902 F.2d 1417, 1423.) In Mayor v. Educational Equality League (1974) 415 U.S. 605 [94 S.Ct. 1323, 39 L.Ed.2d 630], the United States Supreme Court found of “no significance” statistics based on a group of 13, “[i]n large part. . . because the number of positions . . . was too small to provide a reliable sample.” {Id. at p. 611 [94 S.Ct. at p. 1329]; see also id. at p. 612 [94 S.Ct. at p. 1329].) A change of only one person “meant an 8% change in racial composition.” {Id. at p. 611 [94 S.Ct. at p. 1329].) Here, the numbers are even smaller. A single change—retaining the 50 year old and laying off the 34 year old—would not have affected Bechtel’s actions towards Guz in the slightest but would have made the statistics show action favoring older persons. Many cases have found no statistical significance with groups larger than six. (E.g., Vaughan v. MetraHealth Companies, Inc., supra, 145 F.3d at p. 203 [“a sample of seven employees ... is too small for reliable analysis”]; Brocklehurst v. PPG Industries, Inc., supra, 123 F.3d at p. 897 [group of 14 is too small]; Fallis v. Kerr-McGee Corp., supra, 944 F.2d at p. 746 [group of nine “is too small to provide reliable statistical results”]; Simpson v. Midland-Ross Corp., supra, 823 F.2d at p. 943 & fn. 7, and cases cited [reliance on a sample of 17 is “suspect”]; Sengupta v. Morrison-Knudsen Co., Inc. (9th Cir. 1986) 804 F.2d 1072, 1076 [group of 28 is too small].)
Second, Bechtel did not systematically replace older persons with substantially younger ones. Of the six in Guz’s group, all but one were in their 40’s (or 50); one was 34. The replacement of a worker with another “substantially younger than the plaintiff’ {O’Connor v. Consolidated Coin Caterers Corp., supra, 517 U.S. at p. 313 [116 S.Ct. at p. 1310]) might look suspicious, but not these actions. Hartley v. Wisconsin Bell, Inc. (7th Cir. 1997) 124 F.3d 887 considered the “question: how much older than a replacement does a plaintiff have to be in order to pass O’Connor’s test?” {Id. at p. 892.) “While we suspect that the answer depends to some extent on the circumstances in a case, we consider a ten-year difference in ages (between the plaintiff and her replacement) to be presumptively ‘substantial’ under O’Connor. In cases where the disparity is less, the plaintiff may still present a triable claim if she directs the court to evidence that her employer considered her age to be significant. In that instance, the issue of age disparity would be less relevant. ...[![]... Ten years is a reasonable threshold establishing a ‘significant’ and ‘substantial’ gap, which is what O’Connor demands. Yet the line we draw is not so bright as to exclude cases where the gap is smaller but evidence nevertheless reveals the employer’s decision to be motivated by the plaintiff’s age.” {Id. at p. 893.)
*377This assessment seems reasonable. In a given case, the plaintiff might be able to show that an age difference of less than 10 years was significant to the employer. For example, if some important contractual right vested at the age of 50, then replacing a 49 year old with someone younger, even if less than 10 years younger, might be replacing a person with someone substantially younger. But no such evidence exists here. Absent any evidence that Bechtel considered the age differences of persons in their 40’s to be significant, I would find them insignificant. Without more, choosing among various persons in their 40’s gives no cause to suspect age discrimination. Specifically, replacing a 49 year old with a 41 year old is not, by itself, replacing a person with someone substantially younger. Only one person in the group in this case—the 34 year old—was substantially younger than Guz.
Third, the members of the group of six had different qualifications and performed different duties. “[A] plaintiff’s statistical evidence must focus on eliminating nondiscriminatory explanations for the disparate treatment by showing disparate treatment between comparable individuals.” (Fallis v. Kerr-McGee Corp., supra, 944 F.2d at p. 746.) “[T]here must be evidence that [those over 40] had positions and performance ratings that were comparable to [those under 40] who were retained.” (Id. at p. 747.) Here, the group was quite disparate. One, for example, was a secretary, who had duties not remotely similar to Guz’s.
Fourth, any slight weight we may give to plaintiffs statistics is negated by the fact that Guz’s own duties were largely assumed by someone older than he. (Maj. opn., ante, at p. 366.) This fact, even if not itself dispositive, eliminates any suspicious inference that may be drawn from Bechtel’s retaining the two youngest of Guz’s group. “[T]he fact that [the employer] replaced [plaintiff] with [an] even older [employee] contradicts [plaintiff’s] claims of discriminatory animus.” (Brocklehurst v. PPG Industries, Inc., supra, 123 F.3d at p. 897.)
Guz also argues that Bechtel did not follow its own fair layoff procedures, and that this circumstance supports his discrimination claim. The argument is factually dubious but even if correct would fail to suggest age discrimination. A mere failure to follow formal internal policies does not support a discrimination claim. In Vaughan, the employer had an “elaborate Downsizing Policy . . . memorialized in a 144-page Downsizing Manual.” (Vaughan v. MetraHealth Companies, Inc., supra, 145 F.3d at p. 200.) “. . . Cooper, who made the decision to discharge [the plaintiff], admitted he was not familiar with the Downsizing Manual, [and] had never read it. . . .” (Ibid.) Thus, the “district court noted . . . various differences between the Downsizing Manual and Cooper’s actual decision-making process.” (Id. at p. 201.) *378But this showing was not sufficient to support the discrimination claim. “The plaintiff must have developed some evidence on which a juror could reasonably base a finding that discrimination motivated the challenged employment action.” (Id. at p. 202.) The employer’s failure “to follow its own Manual . . . does not even hint that the real motive was age discrimination. ‘The mere fact that an employer failed to follow its own internal procedures does not necessarily suggest that the employer was motivated by illegal discriminatory intent.’ Randle v. City of Aurora, 69 F.3d 441, 454 (10th Cir. 1995). Federal courts cannot ensure that business decisions are always informed or even methodical.” (Id. at p. 203; see also Rose v. Wells Fargo & Co., supra, 902 F.2d at p. 1422; Moore v. Eli Lilly & Co. (5th Cir. 1993) 990 F.2d 812, 819.)
Guz’s inability to present any credible evidence to establish his age discrimination claim supports the superior court’s grant of summary judgment in Bechtel’s favor.
Brown, J., concurred.
Much of the concurring and dissenting opinion is irrelevant to the majority’s holding. The majority does not suggest that age discrimination is lawful; it clearly is not. (Gov. Code, § 12941, subd. (a); see maj. opn., ante, at p. 353, fn. 19.) The majority merely holds that the evidence in this case does not suggest Bechtel engaged in age discrimination. For example, the discussion of Marks v. Loral Corp. (1997) 57 Cal.App.4th 30 [68 Cal.Rptr.2d 1] (conc. & dis. opn. of Kennard, post, at p. 382) is utterly irrelevant. Bechtel never asserted a right to discharge older workers in favor of lower-salaried younger workers to save salaries.
Contrary to the implication of the concurring and dissenting opinion (conc. & dis. opn. of Kennard, post, at p. 385, fn. 4), I do not suggest evidence of age discrimination must be direct, rather than circumstantial. My point is that Guz has presented no meaningful evidence whatsoever, direct or circumstantial, suggesting age discrimination.