CUMMINGS, Chief Judge.

On June 1, 1982, the Supreme Court, ___ U.S. ___, 102 S.Ct. 2288, 73 L.Ed.2d 1297, issued an order in this case granting the petition for a writ of certiorari, vacating the judgment of this Court in the above-titled cause ( 657 F.2d 909, 7th Cir.) and remanding the case for further consideration in light of Kremer v. Chemical Construction Corp., 456 U.S. ___, 102 S.Ct. 1883, 72 L.Ed.2d 262 (1982). We have given careful consideration to the Kremer opinion and the statements filed by the parties pursuant to Circuit Rule 19. Defendant emphasizes the similarity of the Kremer and Unger cases and urges us to dismiss the cases as res judicata. Plaintiff argues that Kremer is distinguishable, or alternatively, that Kremer should not be applied retrospectively. For the reasons articulated below, we have come to the conclusion that this case is controlled by Kremer and that Kremer must be applied retrospectively. Accordingly, we remand the case to the district court with instructions to dismiss the complaint.

I.

The facts of the case are fully set forth in our prior opinion; a brief summary will suffice here. In October 1972, Trudy Unger filed a complaint before the Illinois Fair Employment Practices Commission (FEPC) alleging that her first discharge from the employ of Sirena Division of Consolidated Foods Corporation was discriminatory and her second retaliatory. After a hearing, the Hearing Examiner found against plaintiff on the first discharge and in her favor on the second. The full Commission, however, in 1975 concluded that Trudy Unger was not a victim of sex discrimination on either count. She then sought administrative review in the Circuit Court of Cook County, which reversed the FEPC. Sometime in 1976, defendant appealed the state decision to the Illinois Appellate Court, and in February 1977, plaintiff filed a Title VII suit in Federal District Court. In 1978, the Illinois Appellate Court reversed the Circuit Court and reinstated the decision of the Commission. Leave to appeal to the Illinois Supreme Court was denied; the Illinois state case thus concluded in defendant's favor.

In 1980 in the District Court defendant argued before Judge Roszkowski that plaintiff Unger's suit was barred by res judicata, but the motion to dismiss was denied on Title VII policy grounds and the federal case proceeded to trial. A bench trial before Judge J. Sam Perry resulted in a verdict and judgment in Trudy Unger's favor. Defendant appealed and we affirmed on both the res judicata issue and on the merits of the case.

Prior to our decision in this case, the Supreme Court granted certiorari in Kremer v. Chemical Construction Corp., 623 F.2d 786 (2d Cir. 1980), a Title VII collateral estoppel case. Soon after this Court's decision, defendant filed a petition for certiorari and moved the Supreme Court to consolidate Unger with Kremer. That motion was denied. On May 17, 1982, the Supreme Court held in Kremer that Title VII does not effect a partial repeal by implication of 28 U.S.C. § 1738. A federal court confronted with a Title VII discrimination complaint must give preclusive effect to a prior state court judgment on the same matter — assuming that the plaintiff previously had a "full and fair opportunity" to litigate.

II.

The ultimate question in this case is whether the prior state court judgment in Unger must be accorded preclusive effect under the collateral estoppel analysis of Kremer. As a threshold matter this Court must consider whether the cases are sufficiently similar that the Kremer analysis may appropriately be applied to Unger. In Kremer, plaintiff filed a complaint before the New York State Division of Human Rights, where, after informal proceedings, his complaint was dismissed for lack of probable cause. That decision was sustained throughout the New York State courts. Only after the end of all state court proceedings did Kremer file a Title VII suit in district court. Kremer is thus a straightforward collateral estoppel case. It is virtually the paradigmatic case in that the only question before the Supreme Court was the preclusive effect of a state court proceeding that concededly was final before the federal suit was filed. Unger is complicated procedurally by the fact that plaintiff conducted parallel suits in state and federal court; the collateral estoppel effect of the state court proceeding thus varied from time to time. Moreover, while Kremer lost consistently throughout state and federal proceedings and no federal trial was held, Unger won at the Illinois Circuit Court level, defendant's collateral estoppel argument was rejected in federal district court, and a federal trial was held on the Title VII complaint.

Despite these differences in the sequencing of the state and federal suits in the two cases, it is our conclusion that Kremer is indeed relevant precedent. When the procedural underbrush of Unger is cleared away, it is apparent that a final state judgment preceded the federal trial. As between two actions pending at the same time, the first of two judgments has preclusive effect on the second. 18 Wright, Miller Cooper, Federal Practice and Procedure § 4404; Restatement (Second) of Judgments, § 14 (res judicata), § 27, Comment 1 (collateral estoppel) (1981). Following this rule and the teaching of Kremer, the district court properly might have granted defendant's timely motion to dismiss the suit on a collateral estoppel ground at the time of the Illinois Appellate Court's reversal.

This Court, on remand, is again in the position of reviewing the propriety of the district court's refusal to dismiss the suit on a collateral estoppel ground. The fact that the case went to trial is itself of no consequence. The denial of the collateral estoppel motion to dismiss is not a collateral order appealable under 28 U.S.C. § 1291; appeal of the district judge's decision had to await the end of the trial.

III.

We now turn to the question of whether under Kremer the Illinois state court decision should bar plaintiff. Three points may be disposed of briefly at the outset. First, according to general preclusion principles, a prior action bars a subsequent action only if the same parties and the same cause of action are involved in both suits. We note that the Illinois prohibition against discrimination in employment, Ill.Rev.Stat. ch. 48, § 853, is at least as broad as that of Title VII, and that plaintiff Unger alleged the same discriminatory conduct in state and federal court; thus the same claims of discrimination as well as the same parties are involved in the state and federal suits. Second, the Illinois proceedings satisfy the due process requirements set forth in Kremer, 102 S.Ct. at 1896-99. At an FEPC hearing the parties may be represented by counsel, testimony is under oath and cross-examination is permitted, the rules of evidence used in Illinois courts apply, and compulsory process is available. See Rules and Regulations of the Illinois Fair Employment Practices Commission (issued pursuant to Ill.Rev.Stat. ch. 48, § 856.05 (1975)). The reviewing court is also empowered to remand the case to the agency for the taking of further evidence. Ill.Rev.Stat. ch. 110, § 275(1)(g). Third, a final judgment of the Illinois Appellate Court on administrative review would be accorded preclusive effect by the Illinois courts. See Lee v. City of Peoria, 685 F.2d 196 (7th Cir. 1982). Section 1738 requires the federal court to accord the state court decision a like effect. If plaintiff Unger is to avoid the reach of Kremer, it must be on some other grounds.

Plaintiff Unger argues that Unger and Kremer are distinguishable in terms of the standards of administrative review applied by the Illinois and New York courts. Her contention is that a prior judgment can have preclusive effect only if it was a decision on the very question presented to the second court. The point is essentially that made by Justice Blackmun in his Kremer dissent, 102 S.Ct. at 1903, in which he argued that the New York Appellate Division "made no finding one way or the other concerning the merits of petitioner's claim" (emphasis in original) because the court was reviewing the agency's decision process rather than directly addressing the question of whether Rubin Kremer had been discriminated against. Thus "since the discrimination claim, not the validity of the state agency's decision, is the issue before the federal court," the state court decision should not be a bar according to his dissent. Id. Unger maintains that the New York standard of review was so stringent as to amount to a decision on the merits (and thus a decision on the issue raised in the Title VII complaint), while the Illinois courts' review, being deferential, was a review of the administrative proceeding and thus did not constitute a finding one way or the other as to whether she was a victim of discrimination.

We do not, however, discern any practical difference in the standards of review applied by the Illinois and New York courts. According to the Illinois Appellate Court in Unger, 18 Ill.Dec. at 117-18, 377 N.E.2d at 270-71, "upon review of the dismissal of a charge by the [FEPC], the scope of review is to ascertain from an adequate record whether the commission's order of dismissal is arbitrary, capricious, or an abuse of discretion." The Kremer Court surveyed the New York standard of review in footnote 21, 102 S.Ct. at 1896, and quoted language from a number of New York decisions which require that an agency dismissal for lack of probable cause be affirmed if it appears "virtually as a matter of law that the complaint lacks merit." This might indicate a stringent standard of review. However, one of the quotations describes the standard as requiring only a "rational basis in the record." It also appears that the Supreme Court majority was picking these quotes selectively in an attempt to counter Justice Blackmun's objection that the New York decision was not on the merits. The full texts of the cases cited in footnote 21 contain numerous references to the standard of review, such as: "there is evidence in the record," "substantial evidence," "arbitrary and capricious or characterized by abuse of discretion" and "clearly unwarranted exercise of discretion." The New York and Illinois courts thus appear to be applying the same deferential standard of administrative review. Therefore, there is no basis for distinguishing Kremer and Unger in terms of the stringency of administrative review.

Further, there are at least two reasons for rejecting any argument directed to the absence of a state court de novo trial or some "decision on the merits" equivalent. First, the Supreme Court has already done so; the Kremer Court was fully aware that the New York state courts were conducting an administrative review of a state agency proceeding. Second, Unger's argument proves too much: by her logic, no state court administrative review decision would have to be accorded preclusive effect by the federal courts regardless of whether that state's courts would do so. That, however, would be clearly inconsistent with Section 1738 which is the statutory version of the full faith and credit clause.

IV.

Plaintiff next argues that if this Court concludes that disposition of her case is governed by Kremer, we nevertheless should decline to apply Kremer retrospectively. We must reject that argument.

If we had initially taken the course followed by the Second Circuit (and adopted by the Supreme Court) and held that under the collateral estoppel doctrine the state court decision bars a Title VII suit, or if Kremer had been decided a few months earlier, Trudy Unger would have no argument that she had been deprived retrospectively of federal vindication of her claims; she would merely have lost. Further, the outcome for two similarly situated plaintiffs should not depend on the happenstance of which case reaches the Supreme Court first. As Justice Douglas put it in an analogous retrospectively case, when the relative order of the two cases "is largely a matter of chance . . . [e]qual justice does not permit a [party's] fate to depend on such a fortuity." United States v. Peltier, 422 U.S. 531, 543, 95 S.Ct. 2313, 2320, 45 L.Ed.2d 374 (dissenting opinion).

Under Chevron Oil v. Huson, 404 U.S. 97, 106-107, 92 S.Ct. 349, 355, 30 L.Ed.2d 296, there are three factors to consider in deciding whether a new Supreme Court decision in a civil case should be applied prospectively only:

1. whether the decision establishes a new principle of law, either by overruling clear past precedent on which litigants may have relied, or by deciding an issue of first impression whose resolution was not clearly foreshadowed;

2. whether retrospective application will further or retard application of the new rule; and

3. whether retrospective application would result in substantial injustice to the parties.

In Chevron Oil the issue was whether a state law statute of limitations should be applied to a tort claim based on an injury which occurred on a drilling rig. Under some ten years' precedent within the Fifth Circuit, the Outer Continental Shelf Lands Act had been construed to apply admiralty law and the doctrine of laches to such claims. Plaintiff filed his claim after the Louisiana statute of limitations had expired, but while the suit could have been brought in admiralty. While his suit was pending, the Supreme Court decided a similar case, Rodrigue v. Aetna Casualty Surety Co., 395 U.S. 352, 89 S.Ct. 1835, 23 L.Ed.2d 360, which held that state law governed such claims. Applying the three criteria noted above, the Supreme Court decided not to apply Rodrigue retrospectively to Huson's claim against Chevron Oil. First, the Rodrigue decision was an abrupt departure from settled prior law, and plaintiff, having no reason to suspect such a change in the law, had justifiably relied on laches. Second, the Supreme Court had switched from admiralty law to state law so that workers on the drilling rigs (and their attorneys) could use a body of law familiar to them from their on-shore state residences. Since the change in interpretation of the statute was designed to aid workers in Huson's position, the application of the rule would not be advanced by giving Rodrigue retrospective effect and leaving Huson in limbo between admiralty and state law. Third, Huson would have been left with no forum for his claim because of the fortuity of the change in the applicable limitations period. When these three criteria are applied to Unger, it is apparent that the same arguments for prospectivity cannot be made.

A. The State of Prior Precedent

1. Overruling clear precedent: Defendant argues that 28 U.S.C. § 1738 constitutes the applicable precedent, and Trudy Unger thus had no basis for thinking that a state court decision would not be accorded preclusive effect. Defendant's Statement of Position at 20. That argument is a little disingenuous; the issue is whether Title VII overrides § 1738. It is on this issue that Trudy Unger claims that Kremer is a sharp departure from clear prior law on which she had justifiably relied, but any clarity she sees is largely the product of wishful thinking. She claims reliance on three sources of precedent: the language of Title VII, Alexander v. Gardner-Denver, 415 U.S. 36, 94 S.Ct. 1011, 39 L.Ed.2d 147, and our decision in Batiste v. Furnco, 503 F.2d 447 (7th Cir. 1974), certiorari denied, 420 U.S. 928, 95 S.Ct. 1127, 43 L.Ed.2d 399 (1975). She cannot seriously argue about the language of Title VII since it is precisely the ambiguities in the statute that the Supreme Court construed in Kremer, and while Gardner Denver and Batiste have language broad enough to support her contention that a prior state court judgment should not preclude a federal Title VII suit, the holding in each case is much narrower.

In Alexander v. Gardner-Denver, the Supreme Court held that an employee's statutory right to a federal trial under Title VII may not be foreclosed by prior submission of a claim to binding arbitration under the nondiscrimination clause of a collective bargaining agreement. No state court proceeding was involved. The facts thus provide no support for the proposition that a state court decision would not bar a federal Title VII suit. There is indeed language in the case that could be so construed, but upon closer examination it too fails to support plaintiff's position. For example, plaintiff cites the Supreme Court's comment that "in general, submission of a claim to one forum does not preclude a later submission to another." 413 U.S. at 47-48, 94 S.Ct. at 1019. This statement appears in the context of a comparison between private arbitration and federal de novo trials; state court decisions are nowhere mentioned. Even more telling, however, is the fact that the Supreme Court in a footnote to this statement provided two examples of the principle: EEOC action is not barred by findings and orders of state or local agencies, and an individual's cause of action is not barred by an EEOC finding of no reasonable cause to believe that Title VII had been violated. 415 U.S. at 48, n. 8, 94 S.Ct. at 1019, n. 8.

We also cannot accept plaintiff's argument on prospectivity based on Batiste v. Furnco. Plaintiff points specifically to our statement in Batiste that:

We must agree with the ruling in Cooper v. Philip Morris, Inc., 464 F.2d 9 (6th Cir. 1972) where the court squarely rejected the application of the doctrines of election of remedies and res judicata to Title VII actions where plaintiffs had litigated their charges to final adjudication in state proceedings.

503 F.2d at 450 (emphasis supplied). Admittedly there is ambiguity on the face of this statement; "proceedings" could refer either to state agency proceedings only or to state court proceedings as well. This Court in Unger chose to adopt the latter interpretation, but that does not dispose of the question of how Trudy Unger justifiably could have interpreted this language in 1975 when she filed suit in the Circuit Court of Cook County.

In Batiste v. Furnco, plaintiff's complaint was dismissed by the FEPC hearing officer. Plaintiff then filed suit in district court under Title VII and § 1981. Subsequently, the full Commission reversed the hearing officer and granted relief. Defendant then appealed to the county Circuit Court, but at the time of the district court decision (and apparently during our consideration as well) the Circuit Court proceeding was "still pending." 503 F.2d at 449. It appears that there was never a state court decision, and indeed the parties argued only about the preclusive effect of the FEPC proceeding (defendant also argued election of remedies). Thus while the language in Batiste could be interpreted as including state court judgments under the rubric "proceedings," nothing in the case had anything to do with state court proceedings. The Court probably did not intend that specific construction of its endorsement of Cooper; if it did, the language can only be characterized as dictum.

The inference that in Batiste we meant to refer only to agency proceedings is strengthened by the fact that we explicitly claimed to be following Cooper v. Philip Morris, supra, on the lack of preclusive effect of state proceedings. In that case, plaintiff prevailed before the Kentucky Commission on Human Rights, and then filed a Title VII suit seeking additional relief. Neither party sought recourse to the Kentucky state courts. The district court granted summary judgment for defendant, finding the Commission judgment to be res judicata. The preclusive effect of the agency proceeding was the only issue before the Sixth Circuit, which reversed. Batiste v. Furnco thus cannot be characterized as "settled precedent" upon which Unger justifiably could have relied for the proposition that state court judgments do not have collateral estoppel effect for purposes of Title VII suits.

2. A case of first impression: Plaintiff Unger has a more plausible argument that Kremer is a case of first impression whose outcome was not clearly foreshadowed. As indicated above, Batiste and Alexander provide no basis for confidence that a plaintiff with an adverse state court judgment would be welcome in federal court. On the other hand, as of 1975 when Trudy Unger filed her state court suit, neither was there any case law saying that such a plaintiff could not bring a Title VII suit. Nevertheless, the Supreme Court's language in Chevron Oil still provides difficulties for plaintiff. The phrase "an issue of first impression whose resolution was not clearly foreshadowed" suggests the presence of an issue on which opinion differs and for which the prediction about which way the Supreme Court will turn is unclear. In 1975, when the state trial court was asked to decide this case, one could have guessed that the question of whether Title VII effected a partial repeal of Section 1738 eventually had to be decided one way or the other, but at that time there was no conflict among the circuits in need of resolution. The issue simply had not arisen. It is hard to argue reliance when plaintiff could have done no better than a coin toss in predicting how the Title VII override issue would come out.

At the time Kremer was decided, the question of the preclusive effect of state court judgments on Title VII suits had been considerably sharpened at the Court of Appeals level. The only square conflict between the circuits, however, was between ourselves in Unger and the Second in Kremer and Sinicropi v. Nassau Co., 601 F.2d 60 (2d Cir.), certiorari denied, 444 U.S. 983, 100 S.Ct. 488, 62 L.Ed.2d 411 (1979). A common fact pattern was present in these three cases; plaintiff had appealed an adverse state agency decision to the state courts and sought federal relief after losing in state court. The Second Circuit in Sinicropi (and the Supreme Court in Kremer) also took pains to point out that it was plaintiff who elected to go to state court. After Sinicropi, Unger might thus have anticipated that election of remedies, if not res judicata or collateral estoppel, would create problems.

A different fact pattern was found in other cases reaching the Court of Appeals at about the same time. In Smouse v. General Electric Company, 626 F.2d 333 (3d Cir. 1980), it was defendant, not plaintiff, who took the case to state court following a victory for plaintiff at the state agency level; the Third Circuit distinguished Sinicropi on this ground, although the panel commented that it would have rejected Sinicropi anyway on policy grounds. Similarly, the Eighth Circuit in Gunther v. Iowa State Mens Reformatory, 612 F.2d 1079 (8th Cir.), certiorari denied, 446 U.S. 966, 100 S.Ct. 2942, 64 L.Ed.2d 825 (1980), conceded the "normal preclusion" issue but concluded that "it is our view that Title VII has created a `special circumstance' which warrants an exception to the normal rules of preclusion. This is especially true where plaintiff did not seek state court review, but was forced to defend by [defendant's] appeal." 612 F.2d at 1085.

Thus from the perspective of either 1975 or 1982, in no case reaching the Court of Appeals to which plaintiffs has drawn our attention had a plaintiff other than Trudy Unger been allowed to proceed with a federal Title VII suit after obtaining state court review of an adverse agency determination.

B. Furtherance of The Rule Adopted

The applicability of the second criterion — policy — is ambiguous. The new rule is simply that as a matter of comity, federal courts must accord preclusive effect to state court judgments; Title VII does not override Section 1738. Retrospective application can only advance this rule. There is no compelling reason for giving only prospective effect to Kremer and allowing one more federal case to proceed to finality in the face of a contrary state decision. On the other hand, the federal trial did proceed and since the duplication in court time has already occurred, that aspect of the preclusion doctrine would not be advanced by retrospective application. However, comity, rather than repose per se, was the main focus of Kremer. Thus it appears that there is no policy reason to apply Kremer prospectively. After Kremer, as well as before, the possibility of an unseemly race to judgment between state and federal courts (as occurred in Unger) remains.

C. Substantial Injustice

On the third criterion — injustice — Unger also does not meet the Chevron Oil test. She was not left without a forum in which to pursue her claims. Rather she received a full and fair hearing before an Illinois agency and judicial review by the Illinois Courts. The essence of her injustice/prospectivity argument is that she obtained inconsistent judgments and would prefer to rest on the one in her favor. In applying Kremer retrospectively, it is indeed unfortunate from Unger's point of view, but that is not the same thing as substantial injustice. In sum, when Unger is tested against the three Chevron Oil criteria, we find that it does not present a case for denying retrospective effect to Kremer.

This conclusion is also fully consistent with the Supreme Court's retrospectively analysis in a recent Fourth Amendment case, United States v. Johnson, ___ U.S. ___, 102 S.Ct. 2579, 73 L.Ed.2d 202 (1982). Writing for the majority, Justice Blackmun decried the unprincipled ad hoc application of retrospectively in recent Supreme Court criminal cases. He identified three principles which should guide the retrospective application of a new decision: First, any policy on retrospectively should be predictable procedurally and not be based on the merits of individual cases; a case-by-case balancing of interests leaves no ground for certainty as to when a case should be applied retrospectively. Second, the theory of judicial review itself requires that the court should apply the law as it is at the time of each case as it appears in turn; the court is not a legislative body that is free to give only prospective effect to its decisions. Third, similarly situated parties should be treated alike, with the ultimate outcome not dependent on the speed with which the cases proceeded through the courts.

Justice Blackmun expressly limited his analysis to Fourth Amendment cases, but his reasoning is certainly relevant in a civil context as well. A decision to deny retrospective effect to Kremer would be based not on a principled application of retrospectivity principles, but on the view that as a matter of policy a Title VII plaintiff deserves a federal trial regardless of the state court outcome. This, however, is the policy choice specifically rejected by the Supreme Court in Kremer.

V.

For the above reasons, it is our conclusion that Kremer controls Unger. In the absence of any persuasive reason for denying retrospective effect to Kremer, we hold that the Unger case should be remanded to the district court with instruction to dismiss the complaint.