Opinion for the Court by Circuit Judge BORK.
Dissenting opinion filed by Senior Circuit Judge SWYGERT.
BORK, Circuit Judge.This case is before this court for the second time. Our previous decision reversed a judgment of acquittal entered by the district court after a jury verdict of guilty against Macio Singleton. The panel majority held that the evidence presented was reliable and sufficient to sustain the conviction. Upon remand the district court granted a new trial, and then, upon Singleton’s motion, suppressed the key evidence. The question we now address is whether our prior decision operated to prevent the district court from suppressing evidence we had held reliable. We hold that our previous decision was binding upon the district court and we reverse the suppression order.
I.
The case arises out of a robbery of a fast food restaurant, the rapid apprehension of appellee Singleton, and an onsite showup identification, all on the evening of September 8, 1975. A complete factual account of these occurrences and the surrounding circumstances is contained in United States v. Singleton, 702 F.2d 1159 (D.C.Cir.1983) (“Singleton /”). We need not repeat the details here. After a positive identification, Singleton was charged with four counts of armed robbery, 22 D.C.Code Ann. §§ 2901, 3202 (1981); one count of possession of an unregistered firearm, 26 U.S.C. § 5861(d) (1976); and one count of possession of a firearm not identified by a serial, number, 26 U.S.C. § 5861(i) (1976).
Released on bond, appellee failed to appear for arraignment and a bench warrant was issued. Appellee remained a fugitive for five and one-half years until his arrest on an unrelated arson charge in February, 1981. Prior to trial, appellee filed a motion to suppress the showup identifications on the ground that they were the product of unduly suggestive procedures and that their admission would deprive him of due process.
*178The district court held a suppression hearing and denied appellee’s motion, finding that the showup procedure was not overly suggestive. After a full trial, the jury found appellee guilty on all counts. At the end of the trial, appellee made a motion for new trial and a motion for judgment of acquittal. The trial judge granted appellee’s motion for judgment of acquittal on the ground that the evidence was insufficient to support a finding of guilt beyond a reasonable doubt. He did so specifically on the grounds that the onsite showup was too suggestive and that the recollections of the witnesses at trial more than five years later conflicted with a photograph of appellee taken after his arrest. The judge did not act on the motion for new trial, presumably considering the question mooted by the grant of the other motion.
On appeal, this court reviewed the fifteen-volume transcript of the case, held that neither of the district court’s grounds were sustainable, found that the evidence was sufficient to support the jury’s findings, and reversed and remanded. On remand, however, the district court granted appellee’s pending motion for a new trial. Appellee then renewed his motion to suppress the showup identification testimony and the judge, this time, granted it upon the ground that its introduction would deprive appellee of due process of law. The suppression order, like his previous judgment of acquittal, was based upon the judge’s finding that the testimony was unreliable because of the suggestiveness of the showup and because of the discrepancies between the photograph of appellee taken shortly after his arrest and the witnesses’ conflicting recollection, more than five years later, of appellee’s clothing and appearance that night. The district judge based this finding in part on the proposition that “people who are robbed never forget it.” Joint Appendix (“J.A.”) at 142. This appeal by the government followed.
II.
The government argues that this court’s opinion in Singleton I is the law of the case and therefore requires reversal of the district court’s order granting appellee’s motion to suppress the showup identification testimony. The doctrine of “law of the case” is a rule of practice whereby “courts generally ... refuse to reopen what has been decided.” Messenger v. Anderson, 225 U.S. 436, 444, 32 S.Ct. 739, 740, 56 L.Ed. 1152 (1912). See also In re Multi-Piece Rim Products Liability Litigation, 653 F.2d 671, 678 (D.C.Cir.1981). In White v. Murtha, 377 F.2d 428 (5th Cir.1967), the court stated that,
[t]he “law of the case” rule is based on the salutary and sound public policy that litigation should come to an end. It is predicated on the premise that ... it would be impossible for an appellate court “to perform its duties satisfactorily and efficiently” and expeditiously “if a question, once considered and decided by it were to be litigated anew in the same case upon any and every subsequent appeal” thereof.
Id. at 431 (footnotes omitted). The doctrine also precludes questions “decided by necessary implication as well as those decided explicitly.” Terrell v. Household Goods Carriers’ Bureau, 494 F.2d 16, 19 (5th Cir.), cert. dismissed, 419 U.S. 987, 95 S.Ct. 246, 42 L.Ed.2d 260 (1974); see also Bouchet v. National Urban League, Inc., 730 F.2d 799, 806 (D.C.Cir.1984).
There is an obvious strong similarity between Singleton I and the present appeal (“Singleton II”). The reliability of the showup evidence was central to the first and is the sole issue being considered now. But the cases are not identical in all respects and appellee urges three points of difference that prevent the first appeal from controlling this appeal: (1) Singleton I explicitly stated that it was not deciding the due process issue that is now before us; (2) the review of the sufficiency of the evidence for conviction in Singleton I included more evidence than the showup identifications that were suppressed here; and (3) the standards of review are different in the two cases. See Supplemental Brief for Appellee. We address these points in turn.
*179A.
The panel majority in Singleton I stated that it was not reviewing the district court’s refusal to suppress the evidence of showup identifications prior to trial and said the due process issue was “not properly before us at this time.” Singleton I, 702 F.2d at 1166 n. 23. That was undoubtedly correct. The question, however, is whether Singleton 7’s disavowal of any intention to rule upon the due process issue precludes the rulings that were made from governing this appeal which turns entirely upon the due process issue. We think not. The Singleton I panel did not focus on the due process question since that was not up for review and, being careful not to foreclose a legal issue not considered, noted that fact. That is no more than routine and praiseworthy judicial caution. In no way does that caveat prevent us from concluding that, in truth, the analysis of Singleton I does answer the due process question. It would be extraordinary, in fact, if we, upon finding that Singleton 7’s analysis of the reliability of the showup evidence had effectively decided every element of the due process issue, should nevertheless consider the district court free to reach a different conclusion about the reliability of the evidence. Indeed, the dissenting judge in Singleton I concluded that the majority’s analysis was “identical” to that used in due process cases and tended to foreclose any due process claim here. See Part II.C, infra. It seems clear, therefore, that the question for us is the substance of Singleton 7’s analysis, not the ultimate legal question to which it was addressed.
B.
More evidence was considered by this-court in Singleton I in holding the evidence sufficient to convict than is involved in reviewing the suppression order here. Singleton 7, for example, considered not only the showup identifications but police testimony that one of the robbers was seen fleeing toward the spot where appellee was stopped by the police. Singleton I, 702 F.2d at 1163-64, 1167. We, on the other hand, are considering only the suppression of the showup testimony. Appellee argues, therefore, that Singleton 7, in theory at least, could have held that the evidence was sufficient even if there had been no evidence of the showup identifications. Thus, that case’s determination that the evidence was sufficient does not necessarily determine that the identifications were reliable. That contention would have more force if Singleton I had viewed the evidence as a whole and simply concluded that there was enough for the jury to convict. But that is not what Singleton I did. The panel majority there analyzed separately the reliability of the same evidence the district court has now ordered suppressed in the second trial. The showup identifications were subjected to that analysis independently of the other evidence. Id. at 1165-66. That was necessary because, as the panel majority recognized, the identification testimony was “the keystone of the Government’s case against appellee.” Id. at 1164. The question before us, therefore, is identical to one already decided in Singleton I. That is easily shown.
In Part C of the Singleton I opinion, the court dealt at length with the concerns of the trial judge about the reliability of the showup identification. The question of reliability turned upon two considerations: the suggestiveness of the showup and the inconsistencies in the witness’ recollections some five and one-half years later. This court dealt with both aspects:
We do not believe ... suggestiveness was a factor which, considered with the conflicting testimony of the witnesses, necessitated or justified a verdict of acquittal. Immediate on-the-scene showup identifications, while recognized as inherently suggestive, have long been upheld by this court. “Balancing all the doubts left by the mysteries of human perception and recognition, it appears that prompt confrontations ... will ‘if anything promote fairness by assuring reliability ____’” Notwithstanding its suggestive elements, the showup at issue here took place under circumstances that would tend to promote reliability.
*180Nor do we find fatal the incongruence in the testimony of the witnesses. There are variations, to be sure, between the recollections of the witnesses as to the robber’s facial appearance and clothing and Singleton’s appearance as shown by the photograph taken on the night of his arrest. Such discrepancies are to be expected, however, especially since the witnesses were recalling an event that happened over five years earlier____ [T]he weight to be given to the testimony of the witnesses is for the jury to determine.
Singleton I, 702 F.2d at 1165-66 (footnotes omitted; citations omitted).
It is thus clear that Singleton I decided the issue of the reliability of the showup identifications independently of the other evidence. The fact that more evidence was considered in Singleton I than we consider here does not, therefore, prevent application of the doctrine of law of the case.
C.
Appellee contends that Singleton I cannot be binding in Singleton II because the standard of review is different in the two cases. Supplemental Brief for Appellee at 10-11. In the prior case, he says, we were required to view the evidence in the light most favorable to the government in determining whether the evidence was sufficient to support the conviction. Here, by contrast, we may reverse the district court only if we can conclude that its determination was clearly erroneous. E.g., United States v. Wellins, 654 F.2d 550, 557 & n. 12 (9th Cir.1981); see also 3 C. Wright, Federal Practice and Procedure § 678, at 805 & n. 24 (1982). One is a de novo review, the other a deferential review. That is, of course, a correct statement of the legal formulations of the standards of review where factual determinations by the district judge are attacked. But that is not what is involved here. Appellant is able to produce an apparent anomaly — that we are required to uphold as not clearly erroneous the district court’s determination that the evidence is unreliable despite the decision of this court that it is reliable — only by confining his arguments to standards of appellate review and ignoring the effect of the law-of-the-case doctrine upon the district court.
The answer to appellant’s conundrum is that we are not reviewing fact findings by the district court. We are asking whether, as a matter of law, the district court had any discretion to view the facts differently than we had already done. We think not. The district court was bound by Singleton F s determination that the showup identifications were reliable. That is true because the legal test applied by Singleton I was the same as that to be applied by the district court in deciding the suppression motion. The movant had the extraordinary burden of showing that the evidence was too unreliable even to be placed before a jury. Placing the burden upon the moving defendant to show that no rational jury could convict is the same as viewing the evidence in the light most favorable to the government.1 If the movant does not carry that severe burden, the weight to which the evidence is entitled is a matter for probing and argument by counsel and for ultimate determination by the jury, a body whose capacities of discernment in matters of this sort must not be underrat*181ed. See Manson v. Brathwaite, 432 U.S. 98, 116, 97 S.Ct. 2243, 2254, 53 L.Ed.2d 140 (1977).
Indeed, the identity of the issue before the Singleton I court and the issue before the district court was recognized by the dissenting judge in Singleton I:
The majority states that it declines to address the due process issue. Maj. op. at 1166 n. 23. However, the court adopts an analysis of the showup testimony identical to that used in due process cases, see e.g., id. at 1165-66 & n. 21, and all of the cases relied upon by the court in its analysis are due process cases, see id. at 1165-66 (citing Russell v. United States, 408 F.2d 1280,1284 (D.C.Cir.) cert. denied, 395 U.S. 928, 89 S.Ct. 1786, 23 L.Ed.2d 245 (1969); Wise v. United States, 383 F.2d 206, 208 (D.C. Cir.1967), cert. denied, 390 U.S. 964, 88 S.Ct. 1069, 19 L.Ed.2d 1164 (1968)); maj. op. át 1166 n. 21 (citing Neil v. Biggers, 409 U.S. 188, 199, 93 S.Ct. 375, 382, 34 L.Ed.2d 401 (1972); Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967)). Moreover, because it finds the showup testimony sufficient to allow a reasonable jury to convict, the majority’s analysis tends to foreclose a due process claim. Otherwise, how could evidence be substantial enough to allow a reasonable jury to convict and yet be so unreliable as to be inadmissible on due process grounds?
Nonetheless, the majority clearly leaves open a subsequent appeal by Mr. Singleton raising the due process issue. When such an appeal is brought, the court will have to face the issue directly. In my opinion, admission of such testimony violated due process, so that a fortiori the identification testimony cannot suffice to allow a reasonable jury to convict.
702 F.2d at 1174 n. 18 (Wright, J., dissenting).
This case is to be distinguished from cases enunciating the standard to be applied upon a motion to suppress evidence on the grounds that it was seized without a warrant and without probable cause. See, e.g., United States v. Finefrock, 668 F.2d 1168 (10th Cir.1982). The legal test there is wholly different. As Chimel v. California, 395 U.S. 752, 762, 89 S.Ct. 2034, 2039, 23 L.Ed.2d 685 (1969), made clear, when a search or seizure is warrantless the government carries the burden of justifying the agent’s actions. The general rule is that a warrant is required and the burden of showing probable .cause for a warrantless search is placed upon the government because one seeking an exemption from the rule must show a need for it. Chimel, 395 U.S. at 762, 89 S.Ct. at 2039. Here, by contrast, the government was seeking no exemption and there was no general rule against the evidence it offered.
Not only were the legal issues considered by Singleton I and, subsequently, by the district court the same, so also were the factors considered. The Singleton I court, as we have noted, explicitly used the factors enumerated in Neil v. Biggers, 409 U.S. 188, 199-200, 93 S.Ct. 375, 382-83, 34 L.Ed.2d 401 (1972) (a due process analysis), to evaluate the reliability, and therefore the sufficiency, of the evidence. Singleton I, 702 F.2d at 1166 & n. 21. Under the Biggers due process analysis the court is required to determine reliability by considering the “totality of the circumstances.” 409 U.S. at 199, 93 S.Ct. at 382. Biggers enumerates five factors to assist the court in making its determination:
[1] the opportunity of the witness to view the criminal at the time of the crime,
[2] the witness’ degree of attention,
[3] the accuracy of the witness’ prior description of the criminal,
[4] the level of certainty demonstrated by the witness at the confrontation, and
[5] the length of time between the crime and the confrontation.
Id. at 199-200, 93 S.Ct. at 382-383.
In Singleton I, this court applied the Biggers test, specifically finding four of *182the five suggested factors present, and concluding that in the totality of the circumstances the showup was conducted in a manner that promoted reliability. Singleton I, 702 F.2d at 1166. The court found that each of the witnesses “had ample opportunity to view the [culprits] at close range in good lighting conditions” for a period of several minutes. Id. at 1164-66 & n. 22. The witnesses’ degree of attention varied, but each had focused direct attention on the robber’s face or his clothing. Id. The witnesses testified at the showup that they were positive that Singleton was the robber. Id. at 1164. Indeed, and perhaps most significant, the showup was conducted within a few minutes after the robbery. Id. at 1166. Factor three was the only one not shown by the police or the prosecution. Absence of only this factor, given the court’s conclusions as to the other factors, is not enough to invalidate the entire showup. See United States v. Hammond, 666 F.2d 435 (9th Cir.1982) (failure to give prior identification not fatal under Biggers factors); see also Bass v. Scully, 562 F.Supp. 905 (S.D.N.Y.1983). Thus in Singleton I a due process analysis was used to determine the sufficiency of the evidence. The court recognized that when it said, “[a]lthough ... Biggers [was a] due process case[ ], we think the factors enumerated in Biggers are relevant in evaluating the weight of identification evidence as well as its admissibility.” 702 F.2d at 1166 n. 21.
The district judge, in his subsequent grant of the motion to suppress, based his decision on the alleged unreliability of the evidence produced at the initial trial. The following reference shows that the district court’s rationale was the same as that behind its grant of a motion of acquittal for lack of sufficient evidence at the previous trial:
THE COURT: ... My problem is in ruling on this motion — and I suspect I can, because it was introduced in the trial. I’m as skeptical about it now as I was when I was reluctant to get it in. And if anything, I’m more certain than ever that it had all the earmarks of impermissible suggestion, and had a very dangerous potential for misidentification.
So I’m going to grant the motion. J.A. at 142-43. The district court also found that the recollection of the witnesses, more than five and one-half years after the incident, was not consistent with a photograph of appellee taken shortly after his arrest. Id. at 129-45. In other words, the district court used the identical rationale to grant the motion to suppress the identification testimony as it did in the prior trial to grant the motion of acquittal. Indeed Singleton, on the motion to suppress, argued “that the same reasoning that lead [sic] the Trial Court to grant his motion for judgment of acquittal, required that he reverse himself on the motion to suppress identification.” Brief of Appellee at 6. But that reasoning was precisely what this court rejected in Singleton I.
When applying the law-of-the-case doctrine, the general rule is that where the evidence on a subsequent appeal is substantially the same as that on the first or preceding appeal, all matters, questions, points, or issues adjudicated on the prior appeal are the law of the case on all subsequent appeals and should not be reconsidered or readjudicated therein. See Quern v. Jordan, 440 U.S. 332, 347 n. 18, 99 S.Ct. 1139, 1148 n. 18, 59 L.Ed.2d 358 (1979); Messenger, 225 U.S. at 444, 32 S.Ct. at 740; Schwartz v. NMS Industries, Inc., 575 F.2d 553, at 554-55 (5th Cir.1978); White, 377 F.2d at 431-32; 18 C. Wright, A. Miller & E. Cooper, Federal Practice and Procedure § 4478 (1981); IB J. Moore, Moore’s Federal Practice 110.404[1] (2d ed. 1984). Since reliability has already been found by the Singleton I court, law-of-the-case principles precluded the district court from reconsidering that evidence a second time.2 Indeed, if the district court could *183redetermine the reliability issue, it would be required to do so pursuant to the Biggers test — the same test used by the Singleton I court. This would merely substitute the opinion of the district court for that of the court of appeals. This is exactly what law-of-the-case principles were developed to avoid.
Singleton I was very carefully considered. The record was painstakingly examined. There was a careful majority opinion and a strong dissent. Only a settled conviction that a manifest injustice would otherwise occur would justify us in departing from the doctrine of law of the case. A careful study of Singleton I, the district court’s reasoning, and the briefs now before us in Singleton II produce no such conviction. We cannot go on litigating and adjudicating the reliability of this same evidence forever.
The decision of the district court is reversed and the case is remanded for proceedings consistent with Singleton I and this opinion.
It is so ordered.
. The dissent suggests that the test of "reliability" has different meanings when admissibility is considered before trial and sufficiency is considered after trial. We know of no decision to that effect and the dissent cites none. Nor is the distinction well founded in reason. When a judge asks whether evidence is reliable he asks whether a reasonable person could believe the evidence. The answer ought to be the same whether the legal context is sufficiency or due process.
The dissent’s further observation that most rules of evidence operate to exclude probative evidence is beside the mark. The reliability determination is precisely a decision about whether the evidence is probative in the sense that it is trustworthy enough to lay before jurors. That determination has already been made by this court in Singleton I.
. Contrary to appellee's suggestion, this case does not present the same question as that presented in United States v. Dovico, 261 F.Supp. 862 (S.D.N.Y.1966). There the trial judge found that he was not bound by a decision rendered in a prior trial that evidence was *183admissible. On appeal, the circuit court held that the decision whether the evidence was admissible was for the trial judge at the new trial to make. 380 F.2d 325, 326 n. 1 (2d Cir.), cert. denied, 389 U.S. 944, 88 S.Ct. 308, 19 L.Ed.2d 302 (1967). See also United States v. Akers, 702 F.2d 1145, 1148 & n. 8 (D.C.Cir.1983).
Unlike Dovico, our case involves a district judge ruling on a question previously determined by the appellate court. Although the district court may be free to redecide issues reopenable at the trial court level, it is not free to redecide issues previously determined by the appellate court. Especially if, as here, no new evidence is presented to the trial judge.