ON REMAND FROM THE SUPREME COURT OF THE UNITED STATES
Before TJOFLAT, Chief Judge, KRAVITCH and COX, Circuit Judges. PER CURIAM:Appellant Lloyd Duest is an inmate on Florida’s death row. Previously, we affirmed the district court’s denial of habeas corpus relief as to Duest’s conviction for first-degree murder, but reversed the court’s denial of relief as to his sentence of death. Duest v. Singletary, 967 F.2d 472, 483 (11th Cir.1992). We set aside Duest’s capital sentence because his jury had based its recommendation of death upon consideration of a prior criminal conviction which later was vacated. Id. at 480. Imposition of the death penalty under such circumstances violates the Eighth and Fourteenth Amendments to the United States Constitution, as interpreted in Johnson v. Mississippi 486 U.S. 578, 108 S.Ct. 1981, 100 L.Ed.2d 575 (1988).1 Furthermore, the State did not show beyond a reasonable doubt that Duest would have received the death penalty absent the Johnson error. Duest, 967 F.2d at 482. As such, we held the violation was not constitutionally excusable under the venerable harmless-error standard of Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967) (holding that conviction or sentence obtained in violation of the Constitution must be vacated unless error was “harmless beyond a reason^ able doubt”).
Subsequently, the Supreme Court of the United States granted certiorari, vacated our judgment, and remanded the case to us for further consideration in light of the Court’s recent decision in Brecht v. Abrahamson, 507 U.S. -, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993). Singletary v. Duest, 507 U.S.-, 113 S.Ct. 1940, 123 L.Ed.2d 647 (1993). Brecht, decided after our opinion vacating Duest’s death sentence, partially supplants Chapman. It sets forth a relaxed harmless-error test when federal courts find constitutional trial error on collateral review of state convictions and sentences.
After review in accordance with the Supreme Court’s mandate, we now reinstate our, earlier opinion, except for Part III.B., the section on harmless error, in lieu of which we substitute the revised analysis below. We conclude that, even under the less stringent harmless-error standard enunciated in Brecht, the Johnson error in this case was not harmless. Accordingly, we once again reverse the denial of habeas relief as to sentencing, vacate Duest’s death sentence, and remand the case to the district court for further proceedings.
*1338III
B. Harmless Error
1.
Brecht v. Abrahamson, 507 U.S. -, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993), adopts as the standard for harmless constitutional error in federal habeas corpus cases the harmless-error standard federal appellate courts use on direct review as to noncon-stitutional error.2 That standard first was articulated nearly a half-century ago in Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946). See Brecht, 507 U.S. at -, 113 S.Ct. at 1714. Under Kotteakos, reversal is appropriate only when error resulted in “actual prejudice” to the defendant. Brecht, 507 U.S. at -, 113 S.Ct. at 1722 (citing United States v. Lane, 474 U.S. 438, 449, 106 S.Ct. 725, 732, 88 L.Ed.2d 814 (1986)); Cumbie v. Singletary, 991 F.2d 715, 724 (11th Cir.1993). The Supreme Court has identified the precise contours of this standard in several cases reaching back to Kotteakos.3
Simply put, “actual prejudice” occurs when constitutional error “ lha[s] substantial and injurious effect or influence in determining the jury’s verdict.’ ” Lane, 474 U.S. at 449, 106 S.Ct. at 732 (quoting Kotteakos, 328 U.S. at 776, 66 S.Ct. at 1253); see Brecht, 507 U.S. at -, 113 S.Ct. at 1722 (same). It does not require a showing that but for the error the jury would.have rendered a verdict in favor of the defendant. Kotteakos, 328 U.S. at 763, 66 S.Ct. at 1247 (admonishing that reviewing court should not “speculate upon probable reconviction”); see Brecht, 607 U.S. at -, 113 S.Ct. at 1724 (Stevens, J., concurring); McKinney v. Rees, 993 F.2d 1378, 1385-86 (9th Cir.1993). Nor is it relevant whether the reviewing court is persuaded that the defendant is guilty. Kotteakos, 328 U.S. at 763, 66 S.Ct. at 1247. “[I]t is not the appellate court’s function to determine guilt or innocence.” Id As Justice Rutledge explained in Kotteakos, the issue is not
were [the jurors] right in their judgment, regardless of the error or its effect upon the verdict. It is rather what effect the error had or reasonably may be taken to have had upon the jury’s decision. The crucial thing is the impact of the thing done wrong on the minds of other men [or women], not on one’s own, in the total setting.
_ [I]f one cannot say, with fair assurance, after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is *1339impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected, by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand.
Id. at 764-65, 66 S.Ct. at 1247-48, quoted in part in Brecht, 507 U.S. at-, 113 S.Ct. at 1724 (Stevens, J., concurring), and in Lane, 474 U.S. at 449, 106 S.Ct. at 732 (citations omitted) (emphasis added). The essential question is: Did the constitutional error “substantially influence” the verdict, or, at least, does a “grave doubt” exist as to whether it did? See, e.g., Bank of Nova Scotia v. United States, 487 U.S. 250, 256, 108 S.Ct. 2369, 2374, 101 L.Ed.2d 228 (1988); United States v. Mechanik, 475 U.S. 66, 78, 106 S.Ct. 938, 945, 89 L.Ed.2d 50 (1986) (O’Connor, J., concurring). If so, then the petitioner is entitled to habeas relief.
2.
Harmless error is a mixed question of law and fact subject to de novo review by this court. Jackson v. Dugger, 931 F.2d 712, 717 (11th Cir.), cert. denied, — U.S.-, 112 S.Ct. 452, 116 L.Ed.2d 470 (1991).4 Our task is to examine the entire record and determine whether, in the full context of Duest’s trial and sentencing hearing, the Johnson error was harmless under the principles discussed above. In our opinion, it was not.
Initially, we note that Duest’s sentencing jury recommended death by the slim margin of 7-5. Under Florida law, a 6-6 split is deemed a recommendation against the death penalty. E.g., Harich v. State, 437 So.2d 1082, 1086 (Fla.1983), cert. denied, 465 U.S. 1051, 104.S.Ct. 1329, 79 L.Ed.2d 724 (1984). The sentencing judge may override the jury’s recommendation of life only if no person reasonably could conclude that life imprisonment is an appropriate punishment for this defendant. Hall v. State, 541 So.2d 1125, 1128 (Fla.1989); Tedder v. State, 322 So.2d 908, 910 (Fla.1975). Given the State’s concession that “plenty” of nonstatutory mitigating evidence was present in this case, see State Postconviction Tr. at 121, an override would not have been proper had the jury recommended life. Thus, habeas relief is warranted in this case if we believe even one of the jurors who voted in favor of the death penalty likely was substantially influenced by the evidence of Duest’s prior conviction.
Several facts persuade us that such a likelihood exists. Most important, Duest’s sentencing jury was permitted to consider evidence that was materially inaccurate; the jury did not merely consider properly-admitted evidence in an improper way. That unconstitutional evidence — a subsequently overturned conviction for armed assault with intent to murder — helped portray Duest to the jury not only as an individual with a propensity for criminal violence, but as a recidivous killer. As we previously wrote: “The implication that Duest had acted with murderous intent in committing armed assault might well have had particular impact on a jury considering whether to recommend life or death.” Duest v. Singletary, 967 F.2d 472, 482 (11th Cir.1992), vacated on other grounds, 507 U.S.-, 113 S.Ct. 1940, 123 L.Ed.2d 647 (1993). Furthermore, the jury requested to see the vacated convictions after having deliberated for some time. This fact “suggests strongly that this particular evidence played a significant role in the jury’s ultimate determination that death was the appropriate sentence.” Id. Although it is unclear how much time elapsed between the jury’s request and its arrival at a recommendation of death, the fact that the sentencing hearing took only one day, when combined with the circumstances surrounding the jury’s decision, further calls into question the reliability of the jury’s recommendation. Id. We recognize that the State established three valid aggravating circumstances. Nevertheless, especially when viewed in the light of the close jury vote and the mitigating circumstances, we believe the factors discussed indicate that at least one juror who *1340recommended the death penalty likely was substantially influenced by the vacated conviction.
In accordance with our earlier disposition of this appeal, we REVERSE the district court’s denial of Duest’s petition for a writ of habeas corpus, VACATE Duest’s sentence of death, and REMAND the case to the district court for further proceedings consistent with this opinion.
. Johnson holds that a sentencing jury's consideration of a vacated conviction as the sole evidence of prior criminal conduct to establish an aggravating circumstance unconstitutionally undermines the "need for reliability in the determination that death is the appropriate punishment." 486 U.S. at 584, 108 S.Ct. at 1986. The State does not dispute that Johnson error occurred in this case.
. The harmless-error standard for constitutional violations in all other situations remains the longstanding test of Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967): whether the State has proved the error was harmless beyond a reasonable doubt.
. The threshold issue in all harmless-error inquiries is whether the violation in question amounts to "trial error” or a “structural defect.” Structural defects are not susceptible to harmless-error analysis; trial errors are. Arizona v. Fulminante, 499 U.S. 279, -, 111 S.Ct. 1246, 1264-65, 113 L.Ed.2d 302 (1991) (Rehnquist, C.J., delivering the opinion of the Court). Trial error "occur[s] during the presentation of the case to the jury, and ... may therefore be quantitatively assessed in the context of other evidence presented in order to determine whether its admission was harmless.” Id. at -, 111 S.Ct. at 1264. Structural defects, by contrast, compromise the entire trial process; they involve deprivations of constitutional protection’s so basic, that in their absence no criminal trial can be deemed reliable, nor any punishment fundamentally fair. Id. at -, 111 S.Ct. at 1265. Examples of trial error include the admission of an involuntary confession, see id. at -, 111 S.Ct. at 1264, and the improper use at trial of a defendant’s silence after receiving Miranda warnings, see Brecht, 507 U.S. at -, 113 S.Ct. at 1717. Typical structural errors are the total deprivation of the right to counsel at trial and the unlawful exclusion of members of the defendant’s race from the grand jury. Fulminante, 499 U.S. at -, 111 S.Ct. at 1265.
The Johnson error in this case unquestionably was a trial error. Like the admission of an involuntary confession, the admission of evidence of a subsequently-vacated conviction is a constitutional error which occurs during the presentation of evidence. Its impact is capable of quantitative assessment in relation to all the evidence at trial in order to determine whether it was harmless. It does not infect the entire trial process like the deprivation of trial counsel or the ■ presence at trial of a biased judge. See Fulminante, 499 U.S. at -, 111 S.Ct. at 1265. Thus, we properly turn to analyze under Brecht whether the error in this case merits habeas relief.
. Hence, we are not bound by the Florida Supreme Court’s determination on state postconviction review that Duest’s Johnson claim was harmless under Chapman. See Duest v. Dugger, 555 So.2d 849, 851 (Fla.1990).