dissenting:
The majority “assume[sj,” for purposes of analysis, that the district court’s original charge was in error, but concludes that its response to the jury’s question about the meaning of the erroneous “motivating factor” instruction was fully curative, and/or that MDI’s counsel waived its objection by acceding to a portion of the court’s supplemental instruction. Because I cannot agree that the defendant hospital either waived its objection to or was not prejudiced by the erroneous instruction, and because I believe it important to explain why the original instruction was erroneous, I respectfully dissent.
I.
Employment discrimination cases involving claims of disparate treatment can be divided into two broad categories: (1) cases in which the plaintiff has “direct” evidence of discrimination, and (2) cases in which the plaintiffs evidence of discrimination is “indirect.” See, e.g., Smith v. F.W. Morse & Co., 76 F.3d 413, 421 (1st Cir.1996).2 Under prevailing law, the proper characterization of a plaintiffs ease as direct or indirect largely determines the manner in which the plaintiff may establish the defendant’s liability.
Where the plaintiff’s proof is indirect (as will typically be the case), the plaintiffs claim normally will be adjudicated in accordance with the familiar rules laid out in the line of cases beginning with McDonnell Douglas Corp. v. Green, 411 U.S. 792, 93 S.Ct. 1817, 36 L.Ed.2d 668 (1973), see F.W. Morse, 76 F.3d at 421, and recited in the panel majori*43ty’s opinion, see ante3 At the time of the enactment of the Civil Rights Act of 1991 (“1991 Act”),4 cases in which a plaintiff had direct evidence of discrimination were governed not by McDonnell Douglas, but by Price Waterhouse v. Hopkins, 490 U.S. 228, 109 S.Ct. 1775, 104 L.Ed.2d 268 (1989). See, e.g., Jackson v. Harvard Univ., 900 F.2d 464, 467 (1st Cir.1990). Under the Price Water-house rale, if a plaintiff introduced direct evidence of discrimination sufficient to demonstrate (by a preponderance of the evidence) that an impermissible consideration (race, gender, etc.) was a motivating factor in the contested job action, the burden of proof then shifted to the defendant to prove that it would have taken the same job action against the plaintiff even absent the impermissible motivating factor. See Jackson, 900 F.2d at 467. If the defendant succeeded in so proving, it avoided liability altogether. See Price Waterhouse, 490 U.S. at 244-45, 109 S.Ct. 1775 (plurality opinion); id. at 261-62, 271, 109 S.Ct. 1775 (O’Connor, J., concurring).5
Against the backdrop of McDonnell Douglas and Price Waterhouse, Congress in 1991 amended Title VII of the Civil Rights Act of 1964 by adding two new subsections. Section- 107(a) of the 1991 Act provided the following:
Except as otherwise provided in this title, an unlawful employment practice is established when the complaining party demonstrates that race, color, religion, sex, or national origin was a motivating factor for any employment practice, even though other factors also motivated the practice.
1991 Act, Pub.L. No. 102-166, § 107(a), 105 Stat. 1071, 1075 (1991) (emphasis added) (codified at 42 U.S.C. § 2000e-2(m)). Section 107(b) further added:
On a claim in which an individual proves a violation under [§ 107(a) ] and a respondent demonstrates that the respondent would have taken the same action in the absence of the impermissible motivating factor, the court—
(i) may grant declaratory relief (except as provided in clause (ii)), and attorney’s fees and costs demonstrated to be directly attributable only to the pursuit of a claim under [§ 107(a) ]; and
(ii) shall not award damages or issue an order requiring admission, reinstatement, hiring, promotion, or payment, described in subparagraph [42 U.S.C. § 2000e-5(g)(2)(A)],
Id., § 107(b), 105 Stat. at 1077 (codified at 42 U.S.C. § 2000e5(g)(2)(B)(i)-(ii)).
The 1991 Act was, in large part, Congress’s response to a series of Supreme Court decisions interpreting various provisions of the federal civil rights laws. See Landgraf v. USI Film Products, 511 U.S. 244, 250, 114 S.Ct. 1483, 128 L.Ed.2d 229 (1994). Section 107 was specifically intended to address — and overturn, in part — the holding of Price Waterhouse. See id.; Tanca v. Nordberg, 98 F.3d 680, 681-82 (1st Cir.1996), cert. denied, — U.S. -, 117 S.Ct. 1253, 137 L.Ed.2d 333 (1997); see also H.R.Rep. No. 102-40(1) (Education and Labor Committee), at 45-47 (1991), reprinted in 1991 U.S.C.C.A.N. 549, 583-86; H.R.Rep. No. 102-40(11) (Judiciary Committee), at 16-19 (1991), reprinted in 1991 U.S.C.C.A.N. 694, 709-12.
Under Section 107, if a plaintiff proves that at least ’one of the motivating factors for the defendant’s employment action was an impermissible one — ie., that the defendant acted, at best, with “mixed motives” — liability under Title VII is established. 42 U.S.C. § 2000e-2(m). Section 107 thus abrogates the holding of Price Waterhouse, insofar as that decision permitted an employer with “mixed motives” to avoid liability altogether by proving that it would have taken the same job action against the plaintiff absent any impermissible motivation. Under Section *44107, such proof by an employer does not provide a basis for avoiding liability, but does preclude any recovery of damages by the plaintiff and also bars certain important forms of injunctive relief, including reinstatement. See 42 U.S.C. § 2000e-5(g)(2)(B)(ii).
While it is clear, then, that Section 107 displaces the rule of Price Waterhouse, it is less clear whether Section 107 should have any application in eases traditionally litigated under McDonnell Douglas. The conceptual problem arises from the fact that the “a motivating factor” rubric of § 2000e-2(m)— which was incorporated in the district court’s instructions here — is directly at odds with the established understanding of the plaintiffs burden of proof in the McDonnell Douglas framework, particularly as delineated in cases like St. Mary’s Honor Ctr. v. Hicks, 509 U.S. 502, 506-07, 113 S.Ct. 2742, 125 L.Ed.2d 407 (1993).
Within the McDonnell Douglas framework, a plaintiff bears the ultimate burden of persuading the trier of fact that the defendant’s proffered reasons for the challenged job action are a pretext for discrimination, i.e., that the “true reason” for the decision was discriminatory. See Hicks, 509 U.S. at 507-08, 515, 517, 113 S.Ct. 2742; Smith v. Stratus Computer, Inc., 40 F.3d 11, 16 (1st Cir.1994), cert. denied, 514 U.S. 1108, 115 5.Ct. 1958, 131 L.Ed.2d 850 (1995); Woods v. Friction Materials, Inc., 30 F.3d 255, 260 (1st Cir.1994). To say that a particular motivation was the “true reason” for an action must mean at least that absent that motivation — that is to say, but for it —the action would not have occurred.6 Bibbs v. Block, 778 F.2d 1318, 1321 (8th Cir.1985) (en banc). Thus, satisfying McDonnell Douglas’s burden of proof must entail persuading the finder of fact that the determinative reason for the adverse job action was not the rationale proffered by the defendant, but was, instead, discrimination. See Loeb v. Textron, Inc., 600 F.2d 1003, 1019 (1st Cir.1979) (interpreting McDonnell Douglas to require plaintiff to prove that discriminatory motive was “the determinative factor” or “but for” cause of adverse job action); see also Hartsell v. Duplex Products, Inc., 123 F.3d 766, 772 (4th Cir.1997); Simon v. City of Youngstown, 73 F.3d 68, 70 (6th Cir.1995); Wilson, 55 F.3d at 130 (quoting Miller v. CIGNA Corp., 47 F.3d 586, 598 (3d Cir.1995) (en banc)); Maguire v. Marquette Univ., 814 F.2d 1213, 1216 (7th Cir.1987). Certainly, proving that discrimination was just one motivating factor among other, nondiscriminatory factors— which is all that is required to prove liability under § 2000e-2(m) — cannot suffice to establish that the discriminatory motive was causally determinative, and hence cannot be enough to establish liability and a right to recovery under McDonnell Douglas7 See Goostree v. Tennessee, 796 F.2d 854, 863 (6th Cir.1986) (“The mere showing that ‘sex was a factor’ rather than a ‘but for’ factor is insufficient to establish liability under Title VII.”).
The incompatibility of § 2000e-2(m) with McDonnell Douglas is further demonstrated by the close relationship between § 2000e-2(m) and § 2000e-5(g)(2)(B). If § 2000e-2(m) applied to every disparate treatment case, then § 2000e-5(g)(2)(B) would be triggered in every such case in which the plaintiff prevailed (since the prevailing plaintiff necessarily would have proved a violation under § 2000e-2(m)). As noted above, § 2000e-5(g)(2)(B) permits a defendant to avoid damages and certain other consequences by demonstrating that it would have “taken the same action” absent a discriminatory motive. 42 U.S.C. § 2000e-5(g)(2)(B). But in cases where a plaintiff has prevailed under McDonnell Douglas, the factfinder necessarily will have found that discrimination was the “true reason” for the defendant’s action. See, e.g., Hicks, 509 U.S. at 507-08, 113 S.Ct. 2742. In such cases, it would be absurd to ask the factfinder also to determine, as § 2000e-5(g)(2)(B) would re*45quire, whether the defendant would have “taken the same action” absent a discriminatory motive.
The most plausible way to resolve the apparent conflict between § 2000e-2(m) and McDonnell Douglas is to limit the application of § 2000e-2(m) to “mixed motive” cases that fit the Price Waterhouse mold (cases in which the plaintiff has “direct evidence” of discrimination), and hold it inapplicable in “pretext” cases governed by McDonnell Douglas. This is a more sensible solution, in any event, than drawing the conclusion that Congress, in enacting § 2000e-2(m), intended effectively to abrogate the central tenets of the McDonnell Douglas canon.
The legislative history of the 1991 Act indicates that Section 107 was specifically intended to address the Supreme Court’s holding in Price Waterhouse with respect to the standards of proof governing liability in “mixed motive” cases. See H.R.Rep. No. 102-40(1), supra, at 48, 1991 U.S.C.C.A.N. at 586 (“Section [107] of the legislation ... overrules one aspect of the Supreme Court’s decision in Price Waterhouse.”); H.R.Rep. No. 102-40(11), supra, at 16-17, 1991 U.S.C.C.A.N. at 709-10 (same); see also 187 Cong. Rec. S15476 (Oct. 30, 1991) (interpretive mem. submitted by Sen. Dole) (same); 137 Cong. Rec. H9529 (Nov. 7, 1991) (interpretive mem. submitted by Rep. Edwards) (same). Dicta in one Supreme Court décision and in two of our own prior decisions support this same conclusion. See Landgraf, 511 U.S. at 251, 114 S.Ct. 1483 (“[Section] 107 responds to Price Waterhouse ... by setting forth standards applicable in ‘mixed motive’ cases.”); Tanca, 98 F.3d at 681-82 (“Congress partially overruled Price Water-house in the 1991 Act by allowing a finding of liability and limited relief to plaintiffs in mixed motive cases.”); Smith v. F.W. Morse & Co., 76 F.3d 413, 419 n. 3 (1st Cir.1996) (“[T]he Price Waterhouse framework for proof of ‘mixed motive’ discrimination ... is somewhat changed under the 1991 Act.”). There is absolutely no indication, on the other hand, that Congress intended the 1991 Act to overturn or even affect any aspect of McDonnell Douglas or its derivatives.8
It would be implausible to assume that Congress, in enacting § 2000e-2(m) to address one part of the Supreme Court’s decision in Price Waterhouse, álso intended, sub silentio, to eradicate the established legal landscape governing the litigation of disparate treatment cases. Cf. Square D Co. v. Niagara Frontier Tariff Bureau, Inc., 476 U.S. 409, 418-22, 106 S.Ct. 1922, 90 L.Ed.2d 413 (1986) (legislation overturning a recent Supreme Court decision-involving one particular application of a longstanding judicially constructed rule could not be read as implicit rejection of the entire framework of the rule applied in the overturned decision). A far more reasonable proposition is that 42 U.S.C. § 2000e-2(m) was meant to describe the standard for establishing liability only in “mixed motive” cases, not in standard “pretext” cases governed by McDonnell Douglas. See Fuller v. Phipps, 67 F.3d 1137, 1143-44 (4th Cir.1995) (“Section 107 was intended to benefit plaintiffs in mixed-motive cases; it has nothing to say about the analysis in pretext cases such as this one.”).9 Under *46this reasoning, a district court errs by giving a jury instruction pursuant to § 2000e-2(m), unless the court determines that the plaintiff has adduced evidence of discrimination sufficient to take the case outside the McDonnell Douglas paradigm, and I would so hold.
Here, neither party has argued — and the district court never ruled — ’that this case is anything but a standard “pretext” case governed by the principles of McDonnell Douglas..10 Yet, the district court, explicitly relying on 42 U.S.C. § 2000e-2(m), instructed the jury that the plaintiffs burden was merely to prove that his gender was “a motivating factor” in MDI’s decision to fire him.11 I would find (as the majority “assumes” without deciding) that MDI’s objection to this instruction was well-founded. By instructing the jury that Carey was entitled to prevail upon a finding that discrimination had been merely “a motivating factor” in MDI’s decision to terminate him, the district court erroneously understated the plaintiffs burden of proof under the governing legal framework.
II.
But instructional error warrants reversal, of course, only if it was prejudicial. See Federico v. Order of St. Benedict, 64 F.3d 1, 3 — 4 (1st Cir.1995). The court’s original “a motivating factor” instruction, standing alone, unquestionably prejudiced the defendant by materially understating the plaintiffs true burden of proof. The majority does not dispute this fact. The majority concludes, however, that the district court effectively cured the error in its original instruction with its response to the jury’s question about the meaning of “motivating factor.” I disagree.
As a preliminary matter, it is important to understand that the court’s so-called “supplemental instruction” was, at best, just that — a supplement to the original instruction, given in response to a question that came from the jury after it had already retired to deliberate. The court at no time instructed the jury that the original “a motivating factor” charge was incorrect, to be disregarded, or in any way problematic. Nor did the court advise the jury that the supplemental instruction, to the extent it might have been read as contradicting the original charge, was to be given precedence. Thus, in assessing the question of prejudice, we must ask not simply whether the supplemental instruction itself provided an accurate statement of the law (a question I would answer in the negative, see infra), but whether that instruction — even if it was correct — was sufficiently overriding and forceful in its import that we could confidently say that the jury applied the correct rule of law, in spite of the error contained in the original instruction. Cf. Arthur D. Little, Inc. v. Dooyang Corp., 147 F.3d 47, 53-54 (1st Cir.1998) (“We assume the jury listens to and follows the judge’s entire charge.” (emphasis added)); Tatro v. Kervin, 41 F.3d 9, 18 (1st Cir.1994) (reversing jury’s verdict where a particular instruction, though “standing alone, was proper,” could have *47prejudiced appellant when “taken together” with certain other problematic language).
In its supplemental instruction, the' court stated to the jury that (a) “the plaintiff ... has the burden of establishing that it is more likely than not that gender was a motivating or determining factor in the firing of the plaintiff by the defendant”; (b) “it is for the jury to determine ... [whether] it is more likely than not that the defendant’s decision to terminate the plaintiff was motivated by gender discrimination”; and (c) “If you find that the plaintiff was discharged for reasons other than his gender you must find for the defendant.” This response, in my view, was not sufficient to obviate the prejudicial effect of the court’s original instruction.
As the majority appears to agree, sentence (a) does not accurately describe the plaintiffs burden of proof under McDonnell Douglas and thus could not have been curative.
Sentence (b), considered in isolation, is not necessarily inconsistent with the established standards governing pretext cases, insofar as the “motivated by” formulation leaves open the critical issue of how much of a causative role discrimination must have played in the defendant’s decisionmaking.’ But because the jury was never told that sentence (b) was to trump any conflicting statements in the original instruction, we cannot realistically assume that the jury did consider that sentence in isolation from the “a motivating factor” charge. To the contrary, it only seems fair to presume that the jury’s understanding of sentence (b) was colored by the original “a motivating factor” instruction, and I do not see how we can discount the possibility that the jury thought it could find that the defendant was “motivated by” discrimination so long as gender animus was “a motivating factor.” In any event, I do not think it possible to conclude that the jury, by virtue of sentence (b), must have understood that it could return a plaintiffs verdict only if it found that discrimination had been determinative in the hospital’s decision to fire Carey.
To a limited extent, sentence (c) might be thought to have lessened the prejudicial effect of the original instruction, by suggesting to the jury that a defendant’s verdict was. in order if the jury could “find” that the plaintiffs discharge was caused, to some unspecified degree, by nondiscriminatory reasons. But this is a far cry from saying that sentence (e) provided the jury with a correct understanding of the law. Keeping in mind, again, that the district court never told the jury to disregard the original “a motivating factor” instruction, it would seem more than likely that the jury understood the instructions, as a whole, to require it to return a verdict in Carey’s favor if it found discrimination to be “a motivating factor” in his termination, unless — per sentence (c) — it “f[ou]nd” that Carey was fired “for reasons other than his gender.” This is, of course, an incorrect understanding of the law: a finding that discrimination was “a motivating factor” is insufficient to satisfy the plaintiffs burden of proof in a pretext case, regardless whether the jury could or could not “find” that there were non-diseriminatory reasons for the plaintiffs discharge.12 See Hicks, 509 U.S. at 523-24, 113 S.Ct. 2742 (“Title VII does not award damages against employers who cannot prove a nondiscriminatory reason for adverse employment action, but only against employers who are proven to have taken adverse employment action by reason of (in the context of the present case) [gender].”).
Thus, sentence (c), like (a) and (b), failed to provide the jury with an adequate recitation of the legal standard governing the plaintiffs burden of. proof, and the supplemental instruction did not cure the erroneous “a motivating factor” instruction.
*48hi.
The question remains whether MDI somehow waived any argument that it was prejudiced by the district court’s “a motivating factor” jury instruction. The majority makes much of the fact that, at a few points in the colloquy following the jury’s question, defense counsel appeared to accede to the giving of the supplemental instruction, and endorsed sentence (e) of that instruction in particular. It must be remembered, however, that this entire discussion was prompted by the jury’s question about the meaning of “motivating factor,” a phrase that had been adopted by the district court only over the defendant’s duly-lodged objection. The colloquy preceding the court’s giving of the supplemental instruction did not provide an apt occasion for MDI to reargue whether it had been proper for the district court to give the “a motivating factor” instruction in the first place.
It is only from this perspective that one can fairly understand what the majority refers to as MDI’s “affirmative[ ] ... approval of the supplemental instruction.” In fact, in the course of finally acceding to the supplemental instruction — which, again, was really a response to the jury’s question about language in the original instruction to which MDI had already objected — MDI expressly reserved its argument that the district court had incorrectly instructed the jury on the plaintiffs burden of proof. This is what MDI’s counsel actually said:
although, you know, our position [is] that it should be a “determining/but for” standard, and I don’t want to waive that argument, I think that this [the court’s proposed response to the jury] is acceptable in light of the court’s prior ruling.13
I can see nothing in this statement that compromised the defendant’s ability to argue, on appeal, that it was prejudiced by the court’s “a motivating factor” instruction.
It is true, as the majority observes, that MDI’s counsel at one point did explicitly endorse sentence (e) of the supplemental instruction as a correct statement of the law. I have tried to explain above why I believe (c) was not a fully accurate statement. It is a bit more difficult to say, however, just what effect MDI’s concession should have. On the one hand, it certainly should bar MDI from arguing that the inclusion of sentence (c) in the supplemental instruction was itself reversible error; but that is not a position that MDI has taken in this appeal. On the other hand, MDI’s endorsement of sentence (c) surely cannot be taken as a waiver of its argument that the “a motivating factor” instruction constituted prejudicial error. MDI’s counsel at no time conceded that the supplemental instruction, with or without sentence (c), was sufficient to cure the asserted error in the original charge. In fact, counsel explicitly noted for the record that the hospital did not want to waive its original objection.
In these circumstances, I see no reason that the concession of MDI’s counsel as to statement (c) should require this court, in assessing whether MDI was prejudiced by the “a motivating factor” charge, to conclude that statement (c) effectively cured that erroneous instruction. A contrary outcome, it seems to me, would create a rule of waiver that imposes on trial counsel a “burden of perfection” as “unrealistic” as the burden that the majority asserts would befall the trial court if we were to call for a new trial in this case.
For all of the reasons stated, I would vacate the judgment and remand this ease for a new trial.
. Direct evidence is often said to be the “eviden-tiary equivalent of a 'smoking gun.’ ” Id.; Ayala-Gerena v. Bristol Myers-Squibb Co., 95 F.3d 86, 95-96 (1st Cir.1996); see, e.g., Wilson v. Susquehanna Township Police Dep't, 55 F.3d 126, 128 (3d Cir.1995) (uncontroverted comments by town official who said he was "of one mind” with actual decisionmaker in stating that he "wouldn’t hire any woman” to fill supervisory position sought by plaintiff); Haynes v. W.C. Caye & Co., 52 F.3d 928, 930-31 (11th Cir.1995) (un-rebutted evidence that decisionmaker believed that "a woman was not competent enough” to do the particular job sought by female plaintiff). Indirect evidence includes any other evidence which, if preponderant, could provide a circumstantial or inferential basis for finding discriminatory animus. See, e.g., Fuller v. Phipps, 67 F.3d 1137, 1143 (4th Cir.1995) (evidence of differential treatment); Josey v. John R. Hollingsworth Corp., 996 F.2d 632, 641 (3d Cir.1993) (evidence of discriminatory atmosphere); Ostrowski v. Atlantic Mut. Ins. Cos., 968 F.2d 171, 182 (2d Cir.1992) (statistical evidence). In some circuits, certain strong forms of circumstantial evidence are treated as functionally equivalent to direct evidence. See, e.g., Walden v. Georgia-Pacific Corp., 126 F.3d 506, 513 (3d Cir.1997); Fields v. New York State Office of Mental Retardation & Developmental Disabilities, 115 F.3d 116, 122 (2d Cir.1997).
. Because of the centrality of the notion of ''pretext” in the McDonnell Douglas framework, discrimination cases involving indirect evidence are sometimes referred to as "pretext cases.”
. Pub.L. No. 102-166, 105 Stat. 1071 (1991).
.Because the applicability of Price 'Waterhouse depends upon a plaintiff’s having direct evidence demonstrating that the defendant acted at best with "mixed motives," cases falling within the Price Waterhouse paradigm (direct evidence cases) are commonly labeled as “mixed motive” cases.
. It would be contradictory to say that a particular motivation was the truly operative reason for an action if the action would have occurred whether or not that motivation existed.
. Notably, even under § 2000e-2(m), proof that discrimination was just one motivating factor among others would not automatically entitle the plaintiff to plenary recovery; such proof would only shift the burden to the defendant to prove that discrimination was not the but for cause of the adverse job action. See 42 U.S.C. § 2000e-5(g)(2)(B); Price Waterhouse, 490 U.S. at 244-45, 109 S.Ct. 1775 (plurality opinion).
. At least one portion of the legislative history attempts to harmonize certain provisions of the 1991 Act (albeit not Section 107 in particular) with McDonnell Douglas. See H.R.Rep. No. 102-40(11), supra, at 7, 1991 U.S.C.C.A.N. at 699-700 (explaining that Section 105, which overturned Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 109 S.Ct. 2115, 104 L.Ed.2d 733 (1989), “is not inconsistent with the allocation of burden of proof in disparate treatment cases, as set forth in McDonnell Douglas ... and Burdine ...."). Another portion of the history suggests that Section 107 was simply intended to “restore the decisional law in effect in many of the federal circuits prior to the decision in Price Water-house," H. Rep. No. 102-40(1), supra, at 48, 1991 U.S.C.C.A.N. at 586, again suggesting that the legislation was not intended to disturb McDonnell Douglas or its progeny.
. Cf. Fields v. New York State Office of Mental Retardation & Developmental Disabilities, 115 F.3d 116, 123-24 (2d Cir.1997) (holding that § 107 of the 1991 Act does not require that every discrimination action be treated as a mixed motive case). But cf. Harris v. Shelby Cty. Bd. of Educ., 99 F.3d 1078, 1083-84 (11th Cir.1996) (apparently applying both McDonnell Douglas and the rule of § 2000e-2(m) in a “pretext” case); Woodson v. Scott Paper Co., 109 F.3d 913, 935 n. 29 (3d Cir.1997) (dicta) (suggesting, without reaching the issue, that "there is considerable force to [the] view” that § 2000e-2(m) should apply to "pretext” and “mixed motive” cases *46alike), cert. denied,-U.S. --, 118 S.Ct. 299, 139 L.Ed.2d 230 (1997).
. Indeed, at an earlier stage of this litigation, the district court, in denying MIDI's motion for summary judgment, stated that Carey’s case "rel[ied] on circumstantial evidence," rather than direct evidence, and held that the case would therefore be governed by McDonnell Douglas. Carey did not dispute this conclusion below and does not now dispute it on appeal. It is worth adding that if, at the commencement of trial, the court was uncertain as to how the plaintiff's case might play out, it could have deferred the mixed-motive/pretext determination until such time as its uncertainty could be dispelled. See Starceski v. Westinghouse Elec. Corp., 54 F.3d 1089, 1098 (3d Cir.1995) (choice between "pretext" and "mixed motive” instruction need not be made at beginning of trial).
. The relevant portion of the charge instructed as follows: "It is Mr. Carey’s burden to prove by a preponderance of the evidence ... that his gender was a motivating factor in MDI's discriminatory [sic] firing of him.... If you find, based on the evidence, that Mr. Carey has not proven that his gender was a motivating factor in MDI’s decision to fire him, then you must render a verdict for MDI, the defendant. If you find, however, based on the evidence that Mr. Carey has proven that his gender was a motivating factor in MDI’s decision to fire him, then you must render a verdict for the plaintiff.” MDI objected to this instruction, arguing that § 2000e-2(m) has no proper application in "pretext" cases governed by McDonnell Douglas.
. Thus, for example, if the jury had concluded that Carey had proved only that discrimination was "a motivating factor” in his discharge, but was not prepared to conclude, one way or the other, whether nondiscriminatory reasons played a true role, the jury should have returned a defendant’s verdict. Yet, under the court’s original instruction, even read in conjunction with sentence (e), the juiy might have thought it was required to return a verdict for Carey based on its conclusion that discrimination had been "a motivating factor," since (by hypothesis) the jury would not have been prepared to conclude that . Carey had been "discharged for reasons other than his gender.”
. By “prior ruling,” the defendant was of course referring to the court’s decision to give the "a motivating factor" instruction in the first place.