WILKINSON, Circuit Judge:
This is an employment discrimination action brought under Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e et seq. (1982). Plaintiff Thomas J. Ross, employed by defendant Communications Satel[357]*357lite Corporation (hereinafter COMSAT) until his termination on August 24,1981, complained of sex discrimination by defendant, 42 U.S.C. § 2000e-2(a), and retaliatory harassment and discharge for filing discrimination charges with the Equal Employment Opportunity Commission (EEOC), 42 U.S.C. § 2000e-3(a). Ross sought damages, reinstatement, attorney’s fees and costs.
The district court granted summary judgment for COMSAT. It held that plaintiff’s retaliatory discharge claim was barred due to the collateral estoppel effect of a Maryland unemployment compensation proceeding involving the same parties, which concluded that Ross had been discharged for misconduct. Plaintiff’s remaining sex discrimination and retaliatory harassment claims were found meritless as well. Ross appeals from the grant of summary judgment to COMSAT on his retaliatory discharge and harassment claims.1 We reverse.
Under controlling Maryland law, ignored by the district court, the Maryland unemployment compensation decision was not entitled to receive collateral estoppel effect in a Title VII action alleging retaliatory discharge. Therefore, we remand to the district court for reconsideration of the propriety of summary judgment under the applicable Title VII standards. Believing a clear explanation necessary of this Circuit’s requirements for a plaintiff to prevail in a retaliation action under Title VII, we rule that to disprove a legitimate nondiscriminatory explanation for adverse action offered by the employer, a plaintiff must show that the alleged retaliation would not have occurred “but for” the protected conduct.
I
Ross, a single white male, was employed by COMSAT in its Research and Development Laboratories in Clarksburg, Maryland, from June 8, 1980 until his termination the next year. He held the position of Materials Engineer in the Materials and Process Department, and was under the supervision of Lawrence R. Sparrow, the department manager, who in turn reported to Dr. Edmund S. Rittner, Director of Physical Sciences.
Plaintiff’s original sex discrimination claim, now no longer pressed, was that he experienced harassment, intimidation and unequal treatment following his rejection of purported sexual advances made by a female technician employed by COMSAT in another department of the Laboratories. Although supervisors attempted to resolve the conflict between them, Ross remained dissatisfied with the proposed work arrangements. In early February 1981, Ross threatened to file charges of professional misconduct and sexual harassment against the technician. Subsequently, in a meeting to discuss Ross’s complaint, Rittner allegedly threatened Ross with discharge because of his attitude. On March 13, 1981, Ross filed his initial sex discrimination charge with the EEOC.
Following the filing of this charge, Ross contends that he experienced harassment by COMSAT superiors. Allegedly, his responsibilities and professional status were reduced, he was denied a performance review and annual salary and benefit increases, an article was printed in COMSAT’s News Digest concerning his EEOC charge, and his supervisor Sparrow provided false information about his employment record to other prospective employers. Sparrow, according to Ross, also told one prospective employer, the Applied Physics Laboratory of Johns Hopkins University, that Ross had filed an EEOC charge. Ross’s performance evaluations declined from generally favorable in December 1980 and March 1981 to unfavorable in June 1981, at least [358]*358in part because of his relations with other employees and superiors. On July 8, 1981, Ross filed a retaliation charge with the EEOC.
There is evidence that Ross threatened to “get” Sparrow and Rittner following his unfavorable June performance evaluation, and that he spent substantial portions of his work day complaining to other employees about COMSAT’s management and discussing his legal actions. About August 18, 1981, two other COMSAT employees spoke about Ross’s EEOC charges with him. During the conversation Ross was asked what he would do if he lost his suit, and Ross responded that he might “blow the place up.” The circumstances and seriousness of the threat are in dispute. The employees, however, reported the conversation, and on August 24, 1981, Rittner informed Ross that his employment was terminated because of unacceptable performance and intimidation of other employees. Ross then amended his second EEOC charge on August 26 to allege retaliatory discharge.
After his termination, Ross filed a claim for unemployment compensation with the Maryland Employment Security Administration (ESA). COMSAT contested his claim, and an ESA Claims Examiner found that Ross had been discharged for misconduct, and was therefore disqualified from receiving unemployment benefits for ten weeks, under Md.Code Ann. art. 95A, § 6(c) (1979). Ross appealed that decision, and a de novo hearing was held before an ESA Appeals Referee. The Referee concluded that the “claimant was discharged for misconduct connected with his work within the meaning of Section 6(c) of the Maryland Unemployment Insurance Law,” and affirmed the denial of benefits in a May 27, 1982 opinion. Most significant for this case, the Appeals Referee specifically determined in his opinion:
It is not found that the claimant was discharged as an act of sexual discrimination on the part of the employer. The evidence in this unemployment insurance case, without commenting upon what may be the result in his District Court case, does not substantiate the claimant’s charge that he was discriminated against on the basis of sex. The employer attempted to insulate the claimant from the source of his problem in the work place resulting from a personal relationship with a co-employee. The claimant did not even attempt to give that arrangement an opportunity to work.
The Appeals Referee does not find that the claimant was fired as an act of retaliation as a result of his complaint to the Maryland OSHA.2 The claimant was discharged because he had become so involved in his causes that he, on the employer’s premises during working hours, was interferring [sic] with co-employees to the point that they felt it necessary to complain to management.
When he introduced a threat of future violence into these conversations, the employer was left with no alternative but to discharge the claimant. The claimant’s actions had they been a present threat to do bodily harm or introduce violence into the work place, would have constituted, without question, gross misconduct. But when one considers the circumstances of the claimant in this case, the elements of deliberateness and willfulness are lacking and his actions are, therefore, not gross misconduct. There is no question, however, but that his badgering of his co-employees to the point that they complained does constitute misconduct and did justify his discharge and his disqualification under Section 6(c) of the Maryland Unemployment Insurance Law.
(footnote added).
Ross unsuccessfully appealed the Appeals Referee’s decision to the ESA Board of Appeals, and then to the Circuit Court for Baltimore City, which affirmed the Board of Appeals’ denial of benefits on [359]*359September 20, 1983. Ross took no further appeals. The Appeals Referee rendered his decision while this action, filed March 3, 1982, was pending.
The district court turned first to the question of plaintiff’s discharge. It specifically held that “the doctrine of collateral estoppel precludes plaintiff from here relitigating the issues (1) whether plaintiff’s termination was based on his sex and (2) whether plaintiff was discharged in retaliation for his having filed charges with the EEOC,” because it had been “conclusively established in other litigation that plaintiff was discharged because of misconduct.” Observing that the issues of sex discrimination and retaliation were properly before the administrative agency, having been raised by plaintiff as alternative explanations for his discharge, the district court ruled that the administrative proceedings offered plaintiff a “full and fair opportunity” to litigate those claims and thus satisfied minimal due process requirements. Thus, the district court declared the following administrative findings binding in this Title YII action:
(1) that plaintiff was not discharged as an act of sexual discrimination on the part of defendant, his employer; (2) that plaintiff was not discriminated against on the basis of his sex; (3) that plaintiff was discharged for misconduct; (4) that, specifically, plaintiff was discharged because he became involved during work hours in his own causes and because he badgered and interfered with co-employees and (5) that defendant had no alternative but to discharge plaintiff when he threatened co-employees with future violence.
After disposing of plaintiff’s discharge claims through collateral estoppel, the district court then turned to plaintiff’s other claims of sex discrimination and retaliatory harassment while still an employee, and ruled all to be without merit. The sex discrimination issues are no longer pressed on appeal, and we need only focus on the alleged acts of retaliatory harassment. Although the district court did not expressly rely on collateral estoppel in this portion of its opinion, its rejection of the alleged instances of retaliatory harassment on the job clearly rested to some extent upon the findings of the Appeals Referee, already held binding by the district court. Thus, responding to plaintiffs contentions that defendant reduced his responsibilities and professional status, and denied him an annual review and salary and benefit increases, the district court stated:
Plaintiff was eventually fired because he became involved during work hours in his own causes, and because he badgered and interfered with co-employees. Moreover, as the Appeals Referee found, plaintiff told two other employees that he would ‘blow away’ someone if he failed to win his EEOC case. Under the circumstances, it was entirely appropriate for defendant to reduce plaintiff’s responsibilities and to deny him salary and benefit increases.
Furthermore, the district court held that even if defendant disseminated information to prospective employers concerning plaintiff’s discharge, plaintiff was entitled to no relief, reasoning as follows:
Assuming that defendant told prospective employers that plaintiff had been discharged for misconduct and that he had interfered with co-employees, such statements would not be actionable because they were clearly true. Defendant had every right to advise prospective employers that plaintiff had been discharged for misconduct and further to explain the nature of the misconduct in question.
The district court failed to make any finding on plaintiff’s allegations that COMSAT had given false information to prospective employers prior to his discharge, and had told one prospective employer that Ross had filed an EEOC charge.3 Addressing plaintiff’s claim that he was retaliated [360]*360against by the publication of an article about his EEOC discrimination charge in the COMSAT News Digest, the district court did not rely on any findings by the Appeals Referee, for there were none. Instead, the district court stated that the article was in no way prejudicial or retaliatory, because it accurately reported the facts.4 Plaintiff now contends that publication of the article violated an EEOC policy of maintaining confidentiality on pending claims, a matter on which the district court made no findings.5
Based upon the foregoing analysis, the district court concluded that defendant was entitled to summary judgment under Fed. R.Civ.P. 56 as to all of plaintiffs claims, ruling that as a matter of law “plaintiff was not fired, harassed or discriminated against because of his sex, and that there was no retaliation of any sort undertaken by defendant against plaintiff because plaintiff had filed a complaint of sex discrimination with the EEOC.”
II
We first address the issue of whether the administrative findings of the Maryland ESA, as affirmed by the Circuit Court for Baltimore City, must be given collateral estoppel effect in this Title VII action. We hold that those administrative determinations do not preclude relitigation of the reason for plaintiffs discharge in federal court. Under the “full faith and credit” mandate of 28 U.S.C. § 1738 (1982), the federal courts are required to give preclusive effect to state court judgments where the state courts themselves would do so. Yet, the district court neglected to examine Maryland case law in its collateral estoppel analysis. Had it done so, it would have recognized, as we do now, that the controlling decision of Maryland's highest court, Cicala v. Disability Review Board for Prince George’s County, 288 Md. 254, 418 A.2d 205 (1980), will not permit preclusive effect to be accorded the findings of the Appeals Referee of the Maryland ESA.
Our analysis rests upon the Supreme Court’s interpretation of 28 U.S.C. § 1738 in Kremer v. Chemical Construction Corp., 456 U.S. 461, 102 S.Ct. 1883, 72 L.Ed.2d 262 (1982). Kremer held that
[361]*361“Section 1738 requires federal courts to give the same preclusive effect to state court judgments that those judgments would be given in the courts of the State from which the judgments emerged.” 456 U.S. at 466, 102 S.Ct. at 1889 (footnote omitted).6 As the Court there observed, “[i]t has long been established that § 1738 does not allow federal courts to employ their own rules of res judicata in determining the effect of state judgments. Rather, it goes beyond the common law and commands a federal court to accept the rules chosen by the State from which the judgment is taken.” 456 U.S. at 481-82, 102 5. Ct. at 1897.7 A federal court may not, consistent with § 1738, elect to grant either less preclusive effect than would the state courts, or more. Marrese v. American Academy of Orthopaedic Surgeons, — U.S.-,---, 105 S.Ct. 1327, 1332-35, 84 L.Ed.2d 274 (1985). See Migra v. Warren City School District Board of Education, 465 U.S. 75, 104 S.Ct. 892, 896-98, 79 L.Ed.2d 56 (majority opinion), 899-90 (White, J., concurring) (1984). Rather, the statute commands that we march in tandem, and our role is to determine what the state courts would do, confronted with an identical situation. State law governs, insofar as state law may be discerned.
Here, having conceded the controlling effect of Kremer, the district court erred in neglecting to examine state law in its collateral estoppel ruling. The court’s opinion is barren of any discussion of Maryland collateral estoppel principles or any citation to Maryland cases. Though Kremer itself accorded preclusive effect in a Title VII action to judicial affirmance of a determination by the New York State Division of Human Rights, the Kremer doctrine cuts both ways and may benefit as well as burden Title VII plaintiffs. Various results may ensue, depending upon which state’s law applies and to which prior state proceedings that law is directed.
The Court of Appeals of Maryland, the state’s highest tribunal, has adopted a narrow view of the preclusive effect of a prior judgment from an administrative agency in Cicala v. Disability Review Board for Prince George’s County, 288 Md. 254, 418 A.2d 205 (1980). Cicala, in our opinion, prevents the federal courts from according collateral estoppel effect to the state unemployment compensation findings adverse to Ross. Cicala presented the issue of whether a policeman’s disability was “service connected” under the terms of the county Police Pension Plan. The Disability Review Board had ruled that it was not “service connected,” although the Workmen’s Compensation Commission had previously determined that the injury was “arising out of and in the course of his employment.” Finding that res judicata principles did not apply to the workmen’s compensation determination, the Court of Appeals affirmed the Board’s denial of benefits.
Cicala establishes that under Maryland law a judicial determination by one administrative agency is not binding on another adjudicator “which is seeking to determine an apparently identical issue under a different statute,” 418 A.2d at 211, so long as there exist “substantial differ[362]*362enees” between the statutes themselves. Id. at 212. In reaching its holding in Cicala, the Court of Appeals observed:
The underlying rationale for such a conclusion is that although the issues before the two administrative agencies may appear to be identical, generally they are not. This is so because different statutes have different legislative histories, purposes, scopes of coverage, language, standards, procedures and policies which may dictate opposite results.
418 A.2d at 211. Thus, the Court of Appeals reasoned that although the issues before the Board and the Commission both involved the' question of whether an individual’s injury was related to employment, they were not identical for res judicata purposes, for the pension plan and the workmen's compensation statute had “different origins, coverage, funding sources, procedures and standards.” Id. at 213.
If the Maryland police disability plan and the workmen’s compensation statute at issue in Cicala were so different as to preclude collateral estoppel, Title VII and the Maryland Unemployment Insurance Law cannot be construed as identical under the stringent analysis Maryland law requires. As in Cicala, the issues in both proceedings appear to be similar. Yet the statutes themselves manifest substantial differences. Each statute has distinct enforcement procedures. See 42 U.S.C. § 2000e-5; Md.Code Ann. art. 95A, § 7 (1979 & Cum.Supp.1984). The purposes of the two statutes are unrelated. As stated by the Supreme Court in Griggs v. Duke Power Co., 401 U.S. 424, 429-30, 91 S.Ct. 849, 852-53, 28 L.Ed.2d 158 (1971), Title VII aims to “achieve equality of employment opportunities and remove barriers that have operated in the past.” By contrast, the Unemployment Insurance Law concerns “persons unemployed through no fault of their own.” Md.Code Ann. art. 95A, § 2 (1979).
This distinction in purposes is reflected in the actual standards of recovery. Under Title VII, 42 U.S.C. § 2000e-3(a), it is an “unlawful employment practice for an employer to discriminate against any [employee] ... because he has opposed any practice made an unlawful employment practice by this subchapter, or because he has made a charge, testified, assisted, or participated in any manner in an investigation, proceeding, or hearing under this subchapter.” Md.Code Ann. art. 95A, § 6(c) requires that an individual be temporarily disqualified for unemployment benefits if discharged “for misconduct connected with his work.” The ESA Board of Appeals has defined “misconduct” as “a transgression of some established rule or policy of the employer, the commission of a forbidden act, a dereliction of duty, or a course of wrongful conduct committed by an employee.” Rogers v. Radio Shack, 271 Md. 126, 314 A.2d 113, 117 (1974). Thus, while a Maryland administrative adjudicator is concerned with forbidden conduct on the part of the employee, Title VII directs the factfinder’s attention to a forbidden motive on the part of the employer.
Such distinctions are meaningful because Maryland law makes them so. Whether the due process clause would permit preclusive effect to be assigned judicially affirmed findings of the Maryland ESA we need not decide. See Kremer, 456 U.S. at 481, 102 S.Ct. at 1897. We are convinced, in light of the foregoing considerations, that the Maryland courts would find that the issue resolved by the Appeals Referee, and that to be decided under Title VII, are not identical for collateral estoppel purposes.8
[363]*363III
In denying preclusive effect to the findings of the Maryland ESA, we do not imply that those proceedings are irrelevant. In assessing the genuineness of the issues presented, the district court will be entitled to consider evidence brought out at the state unemployment compensation proceedings, but not to consider itself precluded by the state adjudicator’s conclusions. Similarly, any EEOC proceedings, while they may be relevant to summary judgment, are no substitute for an independent judgment on the part of the district court. See Kremer, 456 U.S. at 470 n. 7, 102 S.Ct. at 1891 n. 7.
In this case we are not satisfied that an independent judgment was made. The district court’s grant of summary judgment for COMSAT rested on the erroneous premise that crucial factual issues were precluded from relitigation by collateral estoppel. The district court relied entirely upon the findings of the ESA Appeals Referee in resolving Ross’s claim of retaliatory discharge. While it is less certain how the district court concluded that Ross’s retaliatory harassment claim was meritless in all aspects, it is apparent that the district court relied heavily on the Appeals Referee’s resolution of the disputed facts.
For example, the district court considered it proper for COMSAT to have reduced Ross’s responsibilities and denied him benefits because he had interfered with ■ co-employees during working hours and threatened violence, citing expressly to the Appeals Referee’s opinion. Further, the district court thought Sparrow’s negative reference given to a prospective employer of Ross “clearly true” because Ross had been discharged for misconduct. Thus, the district court committed reversible error, for its findings embodied the assumption that the earlier Maryland procedure was preclusive. An independent examination of the record to determine whether the evidence reveals “a genuine issue as to any material fact,” Fed.R.Civ.P. 56(c), was required.
In our opinion, the sound course is to reverse the summary judgment and remand this case to the district court for reconsideration of the propriety of summary judgment on each aspect of plaintiff’s retaliatory discharge and harassment claims, and for trial should the evidence on any claim disclose a genuine issue of material fact. It is true that summary judgment presents a legal question, whether “the moving party is entitled to a judgment as a matter of law.” Fed.R.Civ.P. 56(c). An appellate court has power to determine independently whether summary judgment may be upheld on an alternative ground where the basis chosen by the district court proves erroneous. See Charbonnages de France v. Smith, 597 F.2d 406, 416 (4th Cir.1979); 10 C. Wright, A. Miller & M. Kane, Federal Practice and Procedure § 2716 at 657-60 (1983). This case, however, involves a complex array of subsidiary claims within the major categories of discharge and harassment, and, given the factual disputes, it is not clear that the district court could have accorded summary judgment on all the questions presented had preclusive effect not been given to the findings of the Appeals Referee. Allowing the district court to determine initially from all the evidence whether any genuine issues of material fact are presented will promote an informed decision, better frame [364]*364the contentions of the parties, and ensure a proper record for review.
IV
As we have elected to remand this case for reconsideration of summary judgment and trial if necessary, we must provide guidance as to the governing law of Title VII retaliation cases within this Circuit. Defendant contends that Ross must prove ultimately that he would not have been discharged or harassed “but for” the filing of his EEOC charges. Plaintiff argues that he need only show that retaliation for filing charges was “in part” a reason for the adverse employer actions. We conclude that the more stringent “but for” test should be followed in this Circuit, rejecting the “in part” formulation as contrary to both precedent and sound policy.
The general principles of summary judgment, though for the most part settled, bear review here both to aid the district court’s task on remand, and because of their relevance to our ultimate inquiry. The burden is on defendant, as the moving party, to demonstrate the absence of any genuine issue of material fact. Adickes v. S.H. Kress & Co., 398 U.S. 144, 157, 90 S.Ct. 1598, 1608, 26 L.Ed.2d 142 (1970). The facts themselves, and the inferences to be drawn from the underlying facts, must be viewed in the light most favorable to plaintiff, as the party opposing the motion. United States v. Diebold, 369 U.S. 654, 655, 82 S.Ct. 993, 994, 8 L.Ed.2d 176 (1962); Poller v. Columbia Broadcasting System, Inc., 368 U.S. 464, 473, 82 S.Ct. 486, 491, 7 L.Ed.2d 458 (1962). Only where it is “perfectly clear that there are no issues in the case” is summary judgment proper, and even where a directed verdict would be proper after hearing the evidence, the district court should not try the case in advance by summary judgment. Pierce v. Ford Motor Co., 190 F.2d 910, 915 (4th Cir.), cert. denied, 342 U.S. 887, 72 S.Ct. 178, 96 L.Ed. 666 (1951). The non-moving party is in a favorable posture, being entitled “to have the credibility of his evidence as forecast assumed, his version of all that is in dispute accepted, all internal conflicts in it resolved favorably to him, the most favorable of possible alternative inferences from it drawn in his behalf; and finally, to be given the benefit of all favorable legal theories invoked by the evidence as considered.” Charbonnages, 597 F.2d at 414.
In determining whether summary judgment may be granted, the district court must perform a dual inquiry into the genuineness and materiality of any purported factual issues. Whether an issue is genuine calls for an examination of the entire record then before the court in the form of pleadings, depositions, answers to interrogatories, admissions on file and affidavits, under Rule 56(c) and (e). Though the burden of proof rests initially with the moving party, when a motion for summary judgment is made and supported as provided in Rule 56, the nonmoving party must produce “specific facts showing that there is a genuine issue for trial,” rather than resting upon the bald assertions of his pleadings. Fed.R.Civ.P. 56(e). See First National Bank of Arizona v. Cities Service Co., 391 U.S. 253, 289, 88 S.Ct. 1575, 1592, 20 L.Ed.2d 569 (1968). Genuineness means that the evidence must create fair doubt; wholly speculative assertions will not suffice. A trial, after all, is not an entitlement. It exists to resolve what reasonable minds would recognize as real factual disputes. See Cole v. Cole, 633 F.2d 1083, 1089 (4th Cir.1980); Atlantic States Construction Co. v. Robert E. Lee & Co., 406 F.2d 827, 829 (4th Cir.1969).
Care is required in deciding whether the evidence presents a genuine issue of motive, for “summary judgment is seldom appropriate in cases wherein particular states of mind are decisive as elements of [a] claim or defense.” Charbonnages, 597 F.2d at 414. Resolution of questions of intent often depends upon “the credibility of the witnesses, which can best be determined by the trier of facts after observation of the demeanor of the witnesses during direct and cross-examination.” Morrison v. Nissan Motor Co., Ltd., 601 F.2d [365]*365139, 141 (4th Cir.1979). We do not suggest, however, that summary judgment has no role in Title VII litigation. “The fact that motive is often the critical issue in employment discrimination cases does not mean that summary judgment is never an appropriate vehicle for resolution.” International Woodworkers of America v. Chesapeake Bay Plywood Corp., 659 F.2d 1259, 1272 (4th Cir.1981) (emphasis in original). Unsupported allegations as to motive do not confer talismanic immunity from Rule 56.
Even where an issue is genuine, it must prove material as well to escape summary judgment, and at this stage the “but for” test becomes relevant. Given the conflicts in the evidence here, it is possible that defendant would be unable to demonstrate beyond dispute at the summary judgment stage that retaliation played no part in the harassment or discharge of Ross, and yet conclusively show that Ross’s conduct was so inexcusable that COMSAT would have been required to discharge him in any event. As the parties have briefed and argued the question of the applicable test in this Circuit, and it is likely that the choice of tests could determine 1) whether some or all of the retaliation issues can be disposed of by summary judgment, or 2) if tried, how they will be resolved, our consideration of the matter is proper. See United States v. Adamson, 665 F.2d 649, 656 n. 19 (5th Cir.1982), cert. denied, — U.S. -, 104 S.Ct. 116, 78 L.Ed.2d 116 (1983).
The sequence of proof and burdens prescribed by McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-05, 93 S.Ct. 1817, 1824-25, 36 L.Ed.2d 668 (1973), and Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-56, 101 S.Ct. 1089, 1093-95, 67 L.Ed.2d 207 (1981) are applicable to retaliation cases under § 2000e-3 as well as to discriminatory treatment claims. Williams v. Boorstin, 663 F.2d 109, 116 (D.C.Cir.1980); Womack v. Munson, 619 F.2d 1292, 1296 (8th Cir. 1980). The employee is initially required to establish a prima facie case of retaliation by a preponderance of the evidence. Such a prima facie case consists of three elements: 1) the employee engaged in protected activity; 2) the employer took adverse employment action against the employee; and 3) a causal connection existed between the protected activity and the adverse action.9 See, e.g., Smalley v. City of Eatonville, 640 F.2d 765, 769 (5th Cir.1981); Womack, 619 F.2d at 1296; Czarnowski v. Desoto, Inc., 518 F.Supp. 1252, 1257 (N.D. 111.1981); Sogluizzo v. Local 817, International Brotherhood of Teamsters, 514 F.Supp. 277, 280 (S.D.N.Y.1981); Kralowec v. Prince George’s County, 503 F.Supp. 985, 1008 (D.Md.1980), affd, 679 F.2d 883, cert. denied, 459 U.S. 872,103 S.Ct. 159, 74 L.Ed.2d 132 (1982). As in disparate treatment cases, the burden of establishing a prima facie retaliation case “is not onerous.” Burdine, 450 U.S. at 253, 101 S.Ct. at 1093. Once a prima facie case has been presented, the employer then has the burden of producing a legitimate nondiscriminatory reason for the adverse action, thereby rebutting the presumption of retaliation raised by the prima facie case. Womack, 619 F.2d at 1296. The employer is not required to prove the absence of a retaliatory motive, but only to raise a “genuine issue of fact,” Burdine, 450 U.S. at 254, 101 S.Ct. at 1094, as to whether retaliation for protected activity occurred. Finally, if the employer produces a legitimate nondiscriminatory explanation, the employee bears the ultimate burden of proving retaliation by demonstrating that the employer’s proffered reason is pretextual. Womack, 619 F.2d at 1296.
For the employee to disprove a legitimate nondiscriminatory explanation for adverse action, the third stage of the Bur-dine analysis, we determine that he must [366]*366show that the adverse action would not have occurred “but for” the protected conduct. We reject the view that Title VII has been violated if retaliation for protected activity was merely “in part” a reason for the adverse action.
The trend among the circuits is clearly in favor of the “but for” standard, or similar formulations. See Kauffman v. Sidereal Corp., 695 F.2d 343, 345 (9th Cir.1982); Smalley, 640 F.2d at 769; Williams, 663 F.2d at 117; Montiero v. Poole Silver Co., 615 F.2d 4, 9 (1st Cir.1980); Womack, 619 F.2d at 1297. But see Cohen v. Fred Meyer, Inc., 686 F.2d 793, 798 (9th Cir.1982). This Circuit has previously expressed a preference for the “but for” test as well. EEOC v. Federal Reserve Bank of Richmond, 698 F.2d 633, 669 (4th Cir.1983), rev’d oh other grounds sub nom. Cooper v. Federal Reserve Bank, — U.S. -, 104 S.Ct. 2794, 81 L.Ed.2d 718 (1984). The “but for” standard serves the salutary function of directing the factfinder’s attention to the predominant reason for an employer’s adverse action. It ensures that no employee will suffer prejudice by exercising his lawful right to protest perceived discrimination.
Under an “in part” test the filing of unfounded discrimination charges with the EEOC might well be encouraged, as the employee would thereby be insulated from adverse action notwithstanding egregious misconduct, if the employer committed the error of giving any consideration to the EEOC charge. Title VII serves the laudable goal of protecting employee access to agencies and courts. It does not shield employees from normal sanctions for misconduct. “It would be incongruous — and certainly not required by law — to give any employee, even one engaged in exemplary efforts to vindicate the law of the land, a stranglehold on a job irrespective of that employee’s material, work-related flaws.” Williams, 663 F.2d at 116-17.
Finally, Congress has not expressed a stronger preference for preventing retaliation under § 2000e-3 than for preventing actual discrimination under § 2000e-2. The prohibition on retaliation exists simply to ensure that employees will not fear to assert their substantive rights, which are the heart of Title VII. Claimed violations of § 2000e-2 are resolved by the “but for” standard. McDonald v. Santa Fe Trail Transportation Co., 427 U.S. 273, 282 n. 10, 96 S.Ct. 2574, 2580 n. 10, 49 L.Ed.2d 493 (1976). In the absence of strong contrary policy considerations, conformity between the provisions of Title VII is to be preferred.
V
In sum, on remand the district court must determine whether all supporting evidence, with the benefit of all conflicting inferences for plaintiff, creates a genuine factual question that “but for” plaintiff’s filing of the charges with the EEOC, he would not have suffered harassment at the hands of his employer or dismissal from his job. If so, summary judgment is precluded and the ease must proceed to trial.
REVERSED AND REMANDED.