dissenting:
For the reasons expressed in the vacated panel majority opinion, 917 F.2d 1355 (4th Cir.1990), I dissent. The district judge correctly ruled that there were genuine issues of material fact which made summary judgment on the basis of qualified immunity inappropriate. In the district court’s words, “... if there was ever a case that cried for a trial, this is it.... On the face of everything I’ve read and everything I’ve heard, this is what juries are for.” That assessment was an eminently correct one made by an experienced and careful trial judge fully aware of the important purposes served by the qualified immunity defense and that those purposes are not fully realized when resolution of the defense is deferred to trial, but also conscientiously aware of the ultimately limiting principle in summary judgment doctrine that made trial necessary here. The resulting denial of summary judgment should be affirmed.
I
I rely on the vacated panel majority opinion as demonstrating the particular material facts relevant to the qualified immunity defense that remain in genuine issue and for its discussion of the “settled law” issue. I add here some more general observations about the danger-exemplified in the majority opinion — that in an excess of zeal to protect police officers from the rigors and “chilling effect” of trial, courts may be tempted to skew summary judgment doctrine in favor of the officers when immuni*971ty turns on the credibility of their accounts of the circumstances that led them to make mistakes of judgment. Heavily credibility — dependent immunity defenses can of course occur in various kinds of § 1983 cases, but they may be most prevalent where the existence of probable cause is determinative. In such cases, substantive immunity doctrine dictates that an objectively assessed state of mind — what a reasonable police officer in the circumstances knew or should have known about and deduced from the relevant events and conditions bearing upon the existence of probable cause — is likely to be wholly determinative of immunity. See Anderson v. Creighton, 483 U.S. 635, 640-41, 107 S.Ct. 3034, 3039-40, 97 L.Ed.2d 523 (1987). For this reason, in these cases more than most, police officers’ accounts of the circumstances that led them into mistaken judgments are likely to be determinative. And because inevitably — liability being disputed — the officer’s account will be favorable to himself, the credibility of that account is crucial. Frequently, given the nature of police investigations, the officer’s account is not subject to direct refutation by anyone, and must therefore be accepted unless it is manifestly implausible in light of the physical facts, or is intrinsically contradictory, or is otherwise drawn in doubt by extrinsic evidence. Compare, e.g., Torchinsky v. Siwinski, 942 F.2d 257 (4th Cir.1991) (officer’s directly unrefuted account of victim’s false identification of assailant determinative of officer’s immunity), with Sevigny v. Dicksey, 846 F.2d 953 (4th Cir.1988) (officer’s erroneous perception shown to be unreasonable by failure to question available eyewitness who would have corrected it).
In these “probable cause” cases, it is therefore likely that the quite human temptation on the part of police officers to engage in post hoc rationalizations and justifications of their mistakes of judgment may be at its peak — for the very reason that direct refutations of their accounts of the basis for their perceptions of probable cause are so rarely possible. If this be even generally so, the only safeguard for claimants confronted with motions for summary judgment on qualified immunity grounds, when supported by such self-serving accounts, is a rigorous adherence by the courts to the rudiments of summary judgment doctrine. That means, very simply, keeping in mind that the burden is on the police officer as movant for summary judgment, and that where, as in this situation, the basis for the motion is an affirmative defense as to which the movant would have the burden of persuasion at trial, the summary judgment burden is a correspondingly heavy one. See Celotex Corp. v. Catrett, 477 U.S. 317, 330-31 & n. 2, 106 S.Ct. 2548, 2556-57 & n. 2, 91 L.Ed.2d 265 (1986).
The procedural burden thus imposed by summary judgment doctrine upon § 1983 defendants claiming qualified immunity obviously pushes against relevant substantive and procedural immunity doctrine. For immunity doctrine is heavily weighted, substantively and procedurally, in favor of the state actor — avowedly to protect her not only from ultimate liability, but from even having to go to trial. See Mitchell v. Forsyth, 472 U.S. 511, 526, 105 S.Ct. 2806, 2815, 86 L.Ed.2d 411 (1985). This is done substantively by providing a defense that is broader — by excusing constitutional violations resulting from reasonable mistakes of judgment — than is the defense on the merits, see Harlow v. Fitzgerald, 457 U.S. 800, 817-18, 102 S.Ct. 2727, 2737-38, 73 L.Ed.2d 396 (1982), and procedurally, by permitting interlocutory review and reversal of trial court judgments rejecting the defense, in order to avoid trial. See Mitchell v. Forsyth, 472 U.S. 511, 105 S.Ct. 2806, 86 L.Ed.2d 411 (1985).
The conflicting thrusts of summary judgment and qualified immunity doctrine on this matter have led some courts to tinker somewhat with summary judgment practice to relieve its burden on defendant-movants. See, e.g., Pueblo Neighborhood Health Centers, Inc. v. Losavio, 847 F.2d 642, 647-49 (10th Cir.1988) (where § 1983 defendant’s subjective intent an element of constitutional claim, heightened burden placed on non-movant to avoid summary judgment motion on qualified immunity *972grounds). But the great majority of courts, including the Supreme Court, see Harlow, 457 U.S. at 816, 102 S.Ct. at 2737, and this court, Turner v. Dammon, 848 F.2d 440, 443 (4th Cir.1988), have held — or simply assumed as an evident proposition— that summary judgment practice is not to be varied, but applied in the regular way in assessing a qualified immunity defense. See Note, Qualified Immunity and the Allocation of Decision-Making Functions Between Judge and Jury, 90 Col.L.Rev. 1045, 1052-53 & n. 49 (1990) (collecting and analyzing decisions).
This means that in this context as generally, despite the desirability, if possible, of avoiding trial where qualified immunity is asserted, this is not always possible. In this context, as generally, if there are genuine issues of material fact respecting the reasonableness of the police officer’s asserted perception that probable cause existed, the defense may not be established by summary judgment, but must go to trial. There it may yet provide the officer a defense broader than that available to him on the merits, but only there may it be found established.
Facially faithful to Supreme Court and Fourth Circuit precedent, the majority seemingly accepts that summary judgment practice should apply in its regular course here. Certainly it nowhere disputes that proposition. But neither does it expressly recognize or demonstrate any sensitivity to the implications of that proposition for the way in which the widely conflicting forecasts of evidence should be assessed here. Instead, drawing heavily and repeatedly on broad policy considerations underlying qualified immunity doctrine, the majority essentially, though without ever saying so, shifts the burden to the plaintiff as non-movant, either resolving conflicting inferences arising from conflicting versions of critical historical facts in favor of the defendants in an exercise of raw factfinding, or simply sweeping aside as immaterial the existence of flat conflict in the evidence as forecast on certain critical issues. With all respect, I think the majority, out of concern that summary judgment doctrine, regularly applied, might thwart the underlying purposes of qualified immunity doctrine, has simply declined here to apply those aspects unfavorable to the defendants with full and proper rigor.
That is a serious suggestion that should be supported. I will focus on two specific instances and put them in the context of a more general assessment.
Absolutely critical to any fair assessment of whether a reasonable police officer in the circumstances would in the end have thought there was probable cause to take Ms. Gooden into custody, is the question whether the apartment neighbor, Cummings, did, as he claimed under oath, tell the police officers in Beck’s apartment before they took Ms. Gooden away, that they “had the wrong person,” identifying a quarreling couple in the apartment just under Ms. Beck’s apartment as the source of the noise of which Ms. Beck complained. Cummings’ version is not directly disputed in the record — the investigating officers only say that they do not recall talking with him — but the record indicates they did later investigate the possibility of a domestic disturbance in the lower level apartment.
On this state of the record, there is at least a genuine issue of fact as to whether things occurred as Cummings asserts. This is a material fact, for if it is as Cummings asserts, it goes critically to the ultimate issue whether a reasonable police officer would at that point have believed, without a follow-up investigation, that he now had probable cause, whether or not he earlier had. It is undisputed that no such investigation then occurred and that only then, assuming the fact to be as asserted, was Gooden taken involuntarily from the apartment.
The majority fairly recites the Cummings episode in its statement of facts, 963-64, and in the process plainly reveals the existence of the factual issue above identified. But that is the last mention of it. It does not figure in the majority’s ensuing legal analysis; there is no effort to dismiss it as not a “genuine” issue or as one not going to a “material” fact. It simply is dis*973regarded, unless it be thought that the majority’s casual observation that “it will nearly always be the case that witnesses to a crime will differ over what occurred,” 965, is intended to dismiss it as not a genuine issue, or as not concerned with a material fact. Neither properly could be said of this issue.
Equally critical to the ultimate issue is the reasonableness of the officers’ asserted basis for ever thinking they had probable cause. This unavoidably turns on which of conflicting versions of Ms. Gooden’s conduct and appearance are accepted as the true facts.
According to Ms. Gooden, no loud screams ever came from her apartment; instead, at the critical times she was quietly going about her business, talking on the telephone to a person she identified to the police officers and who could have verified her statement if the officers had sought verification. When the officers came into her apartment, she was not disheveled or in any emotional state suggesting recent behavior violent enough to shake the floor and walls of her apartment.
According to the police officers, the loud screams seemed unmistakably to come from Ms. Gooden’s apartment as the complainant Ms. Beck suggested to them, and Ms. Gooden’s appearance was such as to confirm that she may indeed have been engaged in recent violent physical behavior involving screams alternately in female and male voices and throwing herself against the walls.
As matters turned out, it is clear that Ms. Gooden’s version of these critical facts is the true one, and that, as the majority concedes, the officers were simply mistaken throughout. The real question, however, is whether their mistake was a reasonable one, or whether it resulted from such irresponsible and unprofessional conduct that it could not be considered reasonable.
The majority’s approach on this critical matter is simply to disregard the fact that if Gooden’s version is the true one, the very credibility of the officers’ version is necessarily drawn in serious issue. Instead, the majority simply accepts at face value everything the police officers say they heard and saw and thought, and concludes that “there is little dispute here about what the officers perceived.” 965.
With all respect, I think the officers’ obviously self-serving version of what they actually perceived — not to say what a reasonable officer in their position should have perceived — could only be accepted as fact by an exercise of raw factfinding on this record. As indicated, that version is disputed by critical aspects of Ms. Gooden’s conflicting version of her conduct and appearance. It is drawn in further issue by the ready ability of another apartment dweller to locate the source of the commotion in another apartment adjoining the Beck apartment. Perhaps most tellingly, though indirectly, it is drawn in issue by the police officers’ bizarre invocation of a complicated psychological concept that surely is beyond their ken — the split personality disorder — to explain their continued belief that Ms. Gooden, though found peaceably alone in her apartment, nevertheless was the source of the recent “bloodcurdling” screams by male and female voices and the violent shaking of walls and ceilings.
The majority’s analysis simply disregards this evidence drawing in issue the essential credibility of the officers’ account of what they actually saw and heard, or should have seen and heard. Any possible discrepancies in that account (though none ever are conceded) are explained away as the result of the general exigencies of police work in responding to fast-moving emergency situations or as understandable decisions, taken under pressure, to err on the side of caution for their own and a citizen’s sake. That is a perfectly accurate description of the type of dangerous emergency situations which beleaguered police officers — particularly in these times — do frequently encounter and which provide legal excuse, in the form of immunity, for their resulting mistakes. The problem with it here is that it is only an accurate description of the situation in issue if the *974officers’ version is a credible one — the very fact in issue.
The majority’s description of the circumstances could turn out to be an essentially true one: of hard-pressed officers, reasonably relying on available evidence, making a reasonable decision after responsibly checking their perceptions with the investigation possible under the exigencies of the moment. But on this record, there is at least an equally plausible, opposite possibility: of police officers acting too quickly, whether from callousness or indifference, or whatever, to accept the suggestions of an obviously upset apartment dweller; then disregarding, again whether from callousness or indifference, or whatever, all the evidence that questioned their first assumptions including a direct suggestion by an eye-witness that they were mistaken, with a proffer of the true facts; then covering up their mistake by shadings of the facts of what they actually saw and thought, and did or did not do to verify their initial assumption.
I stand on my assessment, and that of the district judge whose decision is now reversed by the en banc court, that summary judgment could be granted here only by failing properly to apply summary judgment principles.
II
The importance of this case — beyond its obvious impact on the litigants — lies in the classic problem it poses of accommodating qualified immunity doctrine’s preference for pre-trial establishment of the defense, with summary judgment’s insistence that, desirable as this may be, it cannot be done if genuine issues of fact material to the defense exist.
This dissent suggests — though with all respect for the difficulty of this problem— that the majority here has succumbed to what may be a rather widespread temptation to put another finger on the scale favoring pretrial establishment of immunity — by skewing summary judgment doctrine. I think this is wrong, and not needed to realize the vital purposes of qualified immunity. More important, it disserves the rights of a § 1983 claimant not to be foreclosed in this way.
Chief Judge Ervin, Judge Murnaghan, Judge Sprouse, and Senior Judge Butzner have asked to be shown as joining in this dissenting opinion.