Vladimir Rodriguez appeals his 109-month sentences for conspiracy to possess with intent to distribute 3, 4-methylenedi-oxymethamphetamine (“MDMA”), also known as ecstasy, in violation of 21 U.S.C. §§ 841(b)(1)(C) and 846, and possession with intent to distribute MDMA, in violation of 18 U.S.C. § 2 and 21 U.S.C. § 841(b)(1)(C). Rodriguez argues that the district court’s calculation of the amount of MDMA attributable to him in determining his base offense level, pursuant to the United States Sentencing Guidelines (“guidelines”), was not supported by the evidence.
Rodriguez also contends for the first time on appeal that the district court violated his Fifth and Sixth Amendment rights by determining his offense level based on facts that were neither charged in his indictment nor proven to a jury. That contention is based on Blakely v. Washington, 542 U.S.-, 124 S.Ct. 2531, 159 L.Ed.2d 403 (2004), which was extended to the federal sentencing guidelines recently in United States v. Booker, 543 U.S. -, 125 S.Ct. 738, 160 L.Ed.2d 621 (2005).
I.
A federal grand jury returned a superseding indictment charging Rodriguez with the drug conspiracy and possession crimes we have described. During his trial on the charges, the government introduced evidence of a conspiracy that in*1293volved Alex Rodriguez, appellant’s brother, supplying MDMA tablets to a number of people in Tampa, Florida, including Jorge Salgado Tergers (“Salgado”), for distribution in Tampa and in other locations. Rodriguez himself was arrested on September 5, 2002, after he transported 2,000 tablets of MDMA from Miami to Tampa as part of a deal between Salgado and a buyer who was working for the government as a confidential informant.
Co-conspirator Salgado, as part of his cooperation with the government, testified during the trial of this case to the following facts. After Salgado traveled from Cuba to the United States in April 2000, he began selling cocaine. When Salgado’s customers requested MDMA tablets, Sal-gado asked Ms friend, Ovidio Gonzalez, to introduce him to Omar Oliva and Alex Rodriguez, two men in Miami who were supplying MDMA. Salgado and Gonzalez then began purchasing MDMA tablets from Oliva and Alex Rodriguez, after which they resold the tablets in Tampa. In conducting these sales, Salgado and Gonzalez either took turns traveling to Miami to pick up the tablets or they employed other people, including Rafael Ruz, to transport them.
Salgado further testified that in 2001 he was arrested for violating the terms of probation that had been imposed by a state court as a result of his previous convictions for dealing MDMA. Thereafter, in October 2001, Salgado was released from custody and traveled to Miami to meet with some MDMA suppliers. Salga-do was introduced to Vladimir Rodriguez, the appellant, who had recently arrived in the United States from Cuba. Salgado described how a few months later Rodriguez joined this conspiracy and began transporting MDMA tablets to Tampa for Salgado and Gonzalez. Salgado later learned that Rodriguez was distributing these drugs inequitably, bringing “like five [thousand tablets] for [Gonzalez]” to sell but only “three [thousand]” for Salgado.
When asked how many times, between Salgado’s release’from custody in October 2001 and Salgado’s arrest in this case on September 5, 2002, Rodriguez had transported MDMA tablets to Tampa for him, Salgado answered: “I couldn’t tell you exactly, but it would have been 10, 12, 13 times during that three-month period.” When asked how many tablets, on average, were involved in each delivery, he answered:
There were three or four thousand— three or four thousand, five thousand, at least to me, for each trip. It could have been the 10 or 12 times. It could have been 25, 30,000 pills, 30, 35,000 pills; I don’t know.
In addition, Salgado testified that, when his girlfriend was arrested for possessing MDMA tablets, he had agreed to work as a confidential informant for Drug Enforcement Administration Special Agent Cesar Diaz. But Salgado continued to deal MDMA tablets illegally while he was working as a Cl, and he ultimately was caught dealing drugs'to two other-people who turned out to be CIs themselves.
Agent Diaz also testified during Rodriguez’s trial. He described how, when he discovered that -Salgado had not ceased his illegal drug activities, their relationship had “sour[ed],” and he began building a criminal case against Salgado. Agent Diaz arranged three controlled buys, using another Cl to act as a purchaser of Salgado’s MDMA tablets on August 2, 8, and 23, 2002, for quantities of 50, 500, and 500 MDMA tablets respectively. Agent Diaz set up a “buy-bust” for September 5, 2002, during which he planned to arrest Salgado.
During his testimony Agent Diaz described how, during that “buy-bust,” Sal-gado entered the home of another Cl, sold *12942,000 tablets, and was arrested. Salgado immediately agreed to cooperate with Agent Diaz and told him that the man in the vehicle outside the home- had transported the tablets. After discovering Rodriguez in the vehicle, law enforcement officials arrested him and recovered from inside the vehicle, among other items: a calculator on Rodriguez’s lap; one notebook without identification that contained entries relating to drug sales; and another notebook - bearing Rodriguez’s name and containing phone numbers, including Oli-va’s number. On cross-examination, Agent Diaz conceded that it was “partially true” that Salgado’s credibility “was shot” with him.
In addition, co-conspirator Ruz testified during this trial. Ruz stated that he had started selling MDMA tablets in Tampa for Salgado in February or March 2001. Although Ruz had begun selling small amounts of MDMA, he “moved up from a hundred pills a week to like 500,” and he “kept getting more and more as the weeks went on,” up until Salgado began sending him to Miami to transport the tablets back to Tampa. Ruz met Rodriguez during his third trip to Miami. After that trip, Salga-do asked Ruz to stop transporting MDMA tablets and to concentrate on sales. Salga-do then had Rodriguez and others transport the tablets. Ruz stated that he personally knew that Rodriguez.had delivered MDMA tablets to Tampa “maybe four times; three, four times,” and that he had met with Rodriguez in a Tampa coffee shop as late as August or September 2002.
Rodriguez took the stand and testified on his own behalf, insisting that he had no involvement in the drug conspiracy. Rodriguez said that he had been waiting in the vehicle in which he was arrested on September 5, 2002 because he wanted Salga-do’s help in bringing Rodriguez’s family to the United States.
The jury convicted Rodriguez on both counts. It was not asked to return any special findings on drug amount.
II.
Prior to sentencing, a probation officer prepared a presentence investigation report (“PSR”), which contained, among other information, a description of the offense conduct, Rodriguez’s personal characteristics, and the fact that he had no known prior convictions. The PSR recommended that Rodriguez’s base offense level be set at 32, pursuant to U.S.S.G. § 2Dl.l(c)(4) (the guideline covering from 1,000 to 3,000 kilograms of marijuana), because Rodriguez’s drug offense involved 30,000 tablets of MDMA, the equivalent of 2,205 kilograms of marijuana.1 It also recommended that the base offense level be adjusted upward two levels, pursuant to U.S.S.G. § 3C1.1, for obstruction of justice, because Rodriguez had falsely testified under oath during his trial that he had no involvement in the crimes for which he was convicted. With a total offense level of 34 and a criminal history category of I, Rodriguez’s resulting guideline range would be 151 to 188 months’ imprisonment.
Rodriguez filed written objections to the PSR’s calculation of the drug amount, contending among other things, that the court should not hold him accountable for 30,000 MDMA tablets because that quantity was supported only by Salgado’s trial testimony, which was inconsistent, uncertain, *1295vague, and therefore unreliable. Rodriguez argued that Salgado’s testimony lacked “sufficient reliability and credibility” because Salgado previously had deceived law enforcement officers with whom he was working as a Cl, and the results of a polygraph examination that Rodriguez had taken contradicted Salgado’s trial testimony.2 In his objections Rodriguez contended that he should be held accountable for only the 2,000 tablets that he had delivered to Salgado on September 5, 2002.
At the sentence hearing, Rodriguez pressed his objection to the PSR’s calculation of drug quantity. Rodriguez argued that: (1) the results of his polygraph examination “impeached” Salgado’s testimony on drug quantity;3 (2) Salgado’s testimony that Rodriguez had made 10 to 12 trips to Tampa and was responsible for 30,000 tablets was vague and uncorroborated; (3) Salgado was a convicted felon who had deceived the government initially while working as a Cl and was not a credible witness; and (4) Ruz had testified that he only knew Rodriguez was in Tampa on three or four occasions.
The government responded that Salgado had no reason to lie about Rodriguez’s relevant conduct because Salgado himself had been held accountable for the same quantity of drugs. The government also pointed out that Rodriguez himself had lied, because his sentencing arguments contradicted his own trial testimony during which he insisted that he had no involvement at all in the drug conspiracy. The government contended that Salgado’s testimony on drug quantity was corroborated by Ruz’s testimony on Rodriguez’s role in the conspiracy. Ruz testified that he had met Rodriguez in Miami and Tampa on several occasions and that Rodriguez was to bring him MDMA tablets. He said that he personally knew Rodriguez had transported MDMA tablets from Miami to Tampa “three, four times.” Finally, the government argued that the results of Rodriguez’s polygraph examination were questionable because his own statements during the examination were inconsistent with his trial testimony.
The district court determined that, although Salgado’s credibility was undermined by his history of dishonesty, Ruz’s testimony was specific as to his interactions with Rodriguez, and Rodriguez’s credibility on the issue of the quantity of drugs he had transported was undermined by his inconsistent trial testimony that he had no involvement in the crime. The court also found that the results of Rodriguez’s polygraph examination were not persuasive because they were inconsistent with his trial testimony and because the questions asked during the examination were not related specifically enough to drug quantity. The court overruled Rodriguez’s objection to the PSR’s calculation of drug quantity.
*1296After granting Rodriguez’s request for a two-level minor role downward adjustment, pursuant to U.S.S.G. § 3B1.2(b), the district court determined .that the resulting guideline range was 97 to 121 months’ imprisonment.4 It sentenced Rodriguez to 109 months’ imprisonment and 3 years’ supervised release.
III.
Rodriguez’s only preserved issue is whether the district court erred in its factual determination of the drug quantity for which Rodriguez was responsible. Our review is limited to clear error. United States v. Auguste, 392 F.3d 1266, 1267 (11th Cir.2004); United States v. Clay, 376 F.3d 1296, 1300 (11th Cir.2004). Where the factfinding resolves a- swearing match of witnesses, the resolution will almost never be clear error. Anderson v. City of Bessemer City, 470 U.S. 564, 575, 105 S.Ct. 1504, 1512, 84 L.Ed.2d 518 (1985) (“[W]hen a trial judge’s finding is based on his decision to credit the testimony of one of two or more witnesses, each of whom has told a coherent and facially plausible story that is not contradicted by extrinsic evidence, that finding, if not internally inconsistent, can virtually never be clear error.”); see also United States v. Gregg, 179 F.3d 1312, 1316 (11th Cir.1999) (“We accord great deference to the district court’s credibility determinations.”).
Section 2D1.1 of the guidelines provides that the base offense level for a possession or a conspiracy drug offense is ordinarily calculated by determining the quantity of drugs attributable to a defendant. See generally U.S.S.G. § 2Dl.l(a). The guidelines also provide that courts shall consider as relevant to this determination “all acts and omissions committed, aided, abetted, counseled, commanded, induced, procured, or willfully caused by the defendant •... that occurred during the commission of the offense of conviction....” U.S.S.G. § lB1.3(a)(l)(A).
When a defendant objects to a factual finding that is used in calculating his guideline sentence, such as drug amount, the government bears the burden of establishing the disputed fact by a preponderance of the evidence. United States v. Agis-Meza, 99 F.3d 1052, 1055 (11th Cir.1996). “The preponderance of evidence is a relaxed evidentiary standard, however, it does not grant the court a license to sentence a defendant in the absence of sufficient evidence when that defendant has properly objected to a factual conclusion.” Id.
“ ‘Where there is no drug seizure or the amount seized does not reflect the scale of the offense, the court shall approximate the quantity of the controlled substance.’” United States v. Frazier, 89 F.3d 1501, 1506 (11th Cir.1996) (quoting U.S.S.G. § 2D1.1, cmt. n.12). Moreover, “[i]n estimating the quantity of drugs attributable to a defendant, a court may base its computation on evidence showing the average frequency and amount of a defendant’s drug sales over a given period of time.” Id. Thus, sentencing “may be based on fair, accurate, and conservative estimates of the quantity of drugs attributable to a defendant.” United States v. Zapata, 139 F.3d 1355, 1359 (11th Cir.1998) (per curiam).
*1297Here, when the government asked Salgado at trial how many times, between Salgado’s release from custody in October 2001 and his arrest in this case on September 5, 2002, Rodriguez had transported drugs for him, Salgado conceded that “I couldn’t tell you exactly[.]” But, when the government asked Salgado how many tablets, on average, were involved in each delivery, Salgado responded:
[t]here were three or four thousand-three or four thousand, five thousand, at least to me, for each trip. It could have been the 10 or 12 times. It could have been 25, 30,000 pills, 30, 35,000 pills; I don’t know.
Salgado also testified that Rodriguez joined the conspiracy a “few months” after October 2001, when Salgado was released from custody, and that Rodriguez began transporting MDMA tablets to Tampa for Salgado and Gonzalez. Salgado testified that, in transporting these tablets, Rodriguez was distributing them inequitably, bringing “like five [thousand] for [Gonzalez]” to sell but only “three [thousand]” for Salgado. After conceding that he could not testify as to the exact number of times that Rodriguez had transported the tablets to Tampa for him, Salgado stated that “it would have been 10,12, 13 times.”
To the extent Rodriguez is arguing that Salgado’s testimony was not' credible because of (1) his criminal history, (2) his admission that he deceived a DEA agent while acting as a Cl, and (3) his desire for the government to recommend a § 5K1.1 departure for substantial assistance, the district court acknowledged those factors, but still found Salgado’s testimony on drug amount to be credible.
Rodriguez contends, in this appeal for the first time, that he was involved only in the “buy-bust” and in the three preceding controlled buys. That contention is inconsistent with his sworn testimony at trial that he had no involvement at all in the drug crimes, and is also inconsistent with his argument at sentencing that he should be held accountable only for the drugs seized as part of the “buy-bust.”
In addition, Salgado’s testimony about Rodriguez’s conduct as a participant in the conspiracy and about the amounts of MDMA tablets involved in the transactions was corroborated by Ruz’s testimony. Ruz met Rodriguez during his third trip to Miami, after which Salgado had Ruz concentrate on sales while Rodriguez and others transported the tablets. Ruz also testified that he knew Rodriguez had delivered MDMA tablets to Tampa “maybe four times; three, four times,” and once within two weeks of Rodriguez’s arrest.
The district court’s calculation of drug amount under the guidelines, therefore, was supported by a preponderance of the evidence and was anything but clearly erroneous.
IY.
Rodriguez’s other contention, the one that he did not preserve with an objection in the district court, is what he refers to in his initial brief as Blakely error, and in his supplemental brief as Booker error. In Blakely, the Supreme Court held that the imposition of a sentencing enhancement under the state of Washington’s mandatory guidelines system based on facts neither admitted by the defendant nor found by the jury violated the defendant’s Sixth Amendment right to trial by jury. 124 S.Ct. at 2534-38. That holding was recently applied to the federal sentencing guidelines in Booker, 125 S.Ct. at 749, where the Supreme Court found “no distinction of constitutional significance between the Federal Sentencing Guidelines and the Washington procedures at issue *1298[in Blakely].” The holding in Booker is that the Sixth Amendment right to trial by jury is violated where under a mandatory guidelines system a sentence is increased because of an enhancement based on facts found by the judge that were neither admitted by the defendant nor found by the jury. See id. at 749-56.
Because Rodriguez did not object on this basis in the district court, our review is only for plain error. See United States v. Cotton, 535 U.S. 625, 631-32, 122 S.Ct. 1781, 1785, 152 L.Ed.2d 860 (2002) (applying plain error review to the defendant’s newly raised claim of error based on Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000)); United States v. Duncan, 381 F.3d 1070, 1073-75 (11th Cir.2004) (reviewing for plain error defendant’s Blakely challenge to the court’s increase of the offense level based on its finding that his offense involved cocaine base, where the defendant’s only objection to this finding at sentencing was based on the sufficiency of the evidence); see also Booker, 125 S.Ct. at 769 (the plain error rule applies and will prevent a new sentencing hearing from being necessary in every case on direct appeal involving a pre-Booker sentence).
Under plain error review, which is authorized by Fed.R.Crim.P. 52(b), federal appellate courts have only “a limited power to correct errors that were forfeited because [they were] not timely raised in [the] district court.” United States v. Olano, 507 U.S. 725, 731, 113 S.Ct. 1770, 1776, 123 L.Ed.2d 508 (1993) (citing Fed.R.Crim.P. 52(b)). An appellate court may not correct an error the defendant failed to raise in the district court unless there is: “(1) error, (2) that is plain, and (3) that affects substantial rights.” Cotton, 535 U.S. at 631, 122 S.Ct. at 1785 (quotations and internal marks omitted). “If all three conditions are met, an appellate court may then exercise its discretion to notice a forfeited error, but only if (4) the error seriously affects the fairness, integrity, or public reputation of judicial proceedings.” Id. (quotations and internal marks omitted).
The Supreme Court has instructed us that plain error review should be exercised .“sparingly,” Jones v. United States, 527 U.S. 373, 389, 119 S.Ct. 2090, 2102, 144 L.Ed.2d 370 (1999), and only “in those circumstances in which a miscarriage of justice would otherwise result.” Olano, 507 U.S. at 736, 113 S.Ct. at 1779 (quotation omitted). In similar fashion we have explained that “our power to review for plain error is ‘limited’ and ‘circumscribed,’ ” United States v. Humphrey, 164 F.3d 585, 588 (11th Cir.1999) (quoting Olano, 507 U.S. at 731, 113 S.Ct. at 1776), that “[t]he plain error test is ‘difficult to meet,’ ” United States v. King, 73 F.3d 1564, 1572 (11th Cir.1996) (quoting United States v. Sorondo, 845 F.2d 945, 948-49 (11th Cir.1988)), and that “[t]he plain error rule places a daunting obstacle before [the appellant].” United States v. Pielago, 135 F.3d 703, 708 (11th Cir.1998).
The first prong of the plain error test is satisfied in this case. Under a mandatory guidelines system, Rodriguez’s sentence was enhanced as a result of findings made by the judge that went beyond the facts admitted by the defendant or found by the jury. The enhancement was based on the judge’s drug quantity findings, and it increased Rodriguez’s base offense level, guidelines range, and sentence under a mandatory guidelines system.5 So, there was Booker error.
*1299The second prong of the plain error test is also met. Although the error was not “plain” at the time of sentencing, “where the law at the time of trial was settled and clearly contrary to the law at the time of appeal — it is enough that the error be ‘plain’ at the time of appellate consideration.” Johnson v. United States, 520 U.S. 461, 468, 117 S.Ct. 1544, 1549, 137 L.Ed.2d 718 (1997). Because of Booker, the error is now plain.
The third prong of the plain error test, however, is another matter. It requires that an error have “affect[ed] substantial rights,” which almost always requires that the error “ ‘must have affected the outcome of the district court proceedings.’ ” Cotton, 535 U.S. at 632, 122 S.Ct. at 1786 (quoting Olano, 507 U.S. at 734, 113 S.Ct. at 1778). The standard for showing that is the familiar reasonable probability of a different result formulation, which means a probability “ ‘sufficient to undermine confidence in the outcome.’ ’’ United States v. Dominguez Benitez, 542 U.S. 74, 124 S.Ct. 2333, 2340, 159 L.Ed.2d 157 (2004) (quoting Strickland v. Washington, 466 U.S. 668, 694, 104 S.Ct. 2052, 2068, 80 L.Ed.2d 674 (1984)). In regard to this third prong, “[i]t is the defendant rather than the [government who bears the burden of persuasion with respect to prejudice.” Olano, 507 U.S. at 734, 113 S.Ct. at 1778.
This burden of showing prejudice to meet the third-prong requirement is anything but easy. The Supreme Court has explained that the .“burden should not be too easy,” because the prejudice standard “should enforce the policies that underpin Rule 52(b) generally, to encourage timely objections and reduce wasteful reversals by demanding strenuous exertion to get relief for unpreserved error.” Dominguez Benitez, 124 S.Ct. at 2340. Particularly important to the present inquiry is what the Supreme Court has said about situations in which it is unclear what effect, if any, an error had.
In Jones, a capital case, the Court was confronted with a situation in which an unobjected to error in the jury instructions may, or it may not, have prejudiced the defendant. In finding that the error did not make it past the third prong of the plain error test, the Court explained:
[E]ven assuming that the jurors were confused over the consequences of deadlock, petitioner cannot show the confusion necessarily worked to his detriment. It is just as likely that the jurors, loath to recommend a lesser sentence, would have compromised on a sentence of life imprisonment as on a death sentence. Where the effect of an alleged error is so uncertain, a defendant cannot meet his burden of showing that the error actually affected his substantial rights. Cf. Romano, 512 U.S. [1], at 14, 114 S.Ct. 2004[, 129 L.Ed.2d 1]. In Romano, we considered a similar argument, namely, that jurors had disregarded a trial judge’s instructions and given undue weight to certain evidence. In rejecting that argument, we noted that, even assuming that the jury disregarded the trial judge’s instructions, “[i]t seems equally plausible that the evidence could have made the jurors more inclined to impose a death sentence, or it could have made them less inclined to do so.” Ibid. Any speculation on the effect of a lesser sentence recommendation, like the evidence in Romano, ivould have had such an indeterminate effect on the *1300outcome of the proceeding that we cannot conclude that any alleged error in the District Court’s instructions affected petitioner’s substantial rights. See Park, 421 U.S. [658], at 676, 95 S.Ct. 1903[, 44 L.Ed.2d 489]; Lopez, 373 U.S. [427], at 436-437, 83 S.Ct. 1381[, 10 L.Ed.2d 462].
Jones, 527 U.S. at 394-95, 119 S.Ct. at 2105 (emphasis added). Cf. Romano v. Oklahoma, 512 U.S. 1, 14, 114 S.Ct. 2004, 2013, 129 L.Ed.2d 1 (1994) (“Either conclusion necessarily rests upon one’s intuition.”). The lesson of the Supreme Court’s Jones decision is that the burden truly is on the defendant to show that the error actually did make a difference: if it is equally plausible that the error worked in favor of the defense, the defendant loses; if the effect of the error is uncertain so that we do not know which, if either, side it helped the defendant loses. Where errors could have cut either way and uncertainty exists, the burden is the decisive factor in the third prong of the plain error test, and the burden is on the defendant.
With these principles in mind, we turn to the task of applying the third prong to this case. The error that was committed in pre-Booker sentencing, like that which occurred in this case, is not that there were extra-verdict enhancements— enhancements based on facts found by the judge that were not admitted by the defendant or established by the jury verdict— that led to an increase in the defendant’s sentence. The error is that there were extra-verdict enhancements used in a mandatory guidelines system. We know for two reasons that if the same extra-verdict enhancements had been found and used in the same way in a non-mandatory guidelines system the result would have been constitutionally permissible.
One reason is that Justice Stevens’ majority opinion in Booker explicitly tells us that: “If the Guidelines as currently written could be read as merely advisory provisions that recommended, rather than required, the selection of particular sentences in response to differing sets of facts, their use would not implicate the Sixth Amendment.” Booker, 125 S.Ct. at 750; id. at 764 (Breyer, J., maj.op.) (“[W]ithout this provision — namely the provision that makes the relevant sentencing rules mandatory and imposes binding requirements on all sentencing judges — the statute falls outside the scope of Apprendi’s requirement.” (internal quotation omitted)).
The other reason we know that the use of extra-verdict enhancements in a non-mandatory guidelines system is not unconstitutional is that the other Booker majority opinion, the one authored by Justice Breyer, specifically provides for extra-verdict enhancements in all future sentenc-ings. Id. at 764-65. The Court specifically held that the guidelines system was constitutional once two parts of the Sentencing Reform Act were excised: the part in 18 U.S.C. § 3553(b)(1) making the guidelines result binding on the sentencing court; and the part in § 3742(e) requiring de novo review of sentences on appeal. Id. at 764 (“With these two sections excised (and statutory cross-references to the two sections consequently invalidated), the remainder of the Act satisfies the Court’s constitutional requirements.”). No part of the Act or guidelines concerning extra-verdict enhancements was excised or modified by the Booker decision. The same extra-verdict enhancement provisions apply after Booker as before.
Stated somewhat differently, if we were to vacate the sentence in this case and remand for resentencing in light of Booker, the district court would be faced with exactly the same evidence presenting exactly the same factual issues that it has *1301already resolved, and it would be required to at least consider exactly the same guideline enhancement provisions it has already applied. No guideline provisions have changed in any way. All that has changed is that the guidelines range is now advisory; it no longer dictates the final sentencing result but instead is an important factor that the sentencing court is to consider along with the factors contained in § 3553(a) in reaching the sentencing result. Id. at 764-65.
The point of all this for third-prong plain error review purposes is that the constitutional error whose prejudicial measure we take is not the use of extra-verdict enhancements. Their use remains a constitutional part of guidelines sentencing in the post -Booker era. The constitutional error is the use of extra-verdict enhancements to reach a guidelines result that is binding on the sentencing judge; the error is the mandatory nature of the guidelines once the guidelines range has been determined. Therefore, in applying the third prong, we ask whether there is a reasonable probability of a different result if the guidelines had been applied in an advisory instead of binding fashion by the sentencing judge in this case. The obvious answer is that we don’t know. If the district court judge in this case had the liberty of increasing or decreasing Rodriguez’s sentence above or below the guidelines range, he might have given Rodriguez a longer sentence, or he might have given a shorter sentence, or he might have given the same sentence. The record provides no reason to believe any result is more likely than the other. We just don’t know.
Because we don’t know, the Supreme Court’s Jones decision is controlling. It tells us that where the effect of an error on the result in the district court is uncertain or indeterminate — where we would have to speculate — the appellant has not met his burden of showing a reasonable probability that the result would have been different but for the error; he has not met his burden of showing prejudice; he has not met his burden of showing that his substantial rights have been affected. Jones, 527 U.S. at 394-95, 119 S.Ct. at 2105. Because this appellant has not carried his burden as to the third prong of the plain error test, we have no occasion to decide how he would have fared under the fourth prong.
We are aware that three other federal courts of appeals have reached results different from our own in regard to whether Booker error, is plain error. We have carefully read their opinions and are unpersuaded by them. The Fourth Circuit’s conclusion in United States v. Hughes, 396 F.3d 374, 2005 WL 147059 (4th Cir. Jan.25, 2005), that the third prong of the plain error test is met in Booker-error cases is based on measuring prejudice in terms of the effect the extra-verdict enhancements in that case had in determining the guidelines sentencing range. Id. at 378-79, 380, *3-4, 5. Because those enhancements increased the sentencing range, they resulted in a higher sentence and in the Fourth Circuit’s view prejudicially affected the outcome. Id. at 380, *5. The problem with the reasoning is that, as we have already discussed, both majority opinions in Booker make clear that the decisive factor that makes pre-Booker sentencing problematic is not extra-verdict enhancements but their use in a mandatory guidelines system. Indeed, all nine justices joined one or both of those opinions, so all nine agreed that the use of extra-verdict enhancements in an advisory guidelines system is not unconstitutional.
Extra-verdict enhancements are to be determined and used in the post-Booker world. Indeed, the Fourth Circuit’s remand instructions to the district court in *1302Hughes recognize as much, because they tell the court to consider the guidelines range as increased by the extra-verdict enhancements when it re-sentences the defendant. Id. at 384-85, *8 (“Because we conclude that the district court correctly determined the range prescribed by the guidelines, on remand the court shall consider that range as well as other relevant factors set forth in the guidelines and those factors set forth in § 3553(a) before imposing the sentence.” (footnote omitted)). If the finding and consideration of the extra-verdict enhancements were the constitutional error in Booker, the Supreme Court in that case would not have permitted, and the Fourth Circuit itself in Hughes would not have directed, district courts to continue to find and consider such enhancements. Extra-verdict enhancements are still to be used. The change is the role of the guidelines range in the sentencing process, and it is that change against which the third-prong prejudice measure must be taken.
Another way of putting it is that the Hughes reasoning is wrong because it disconnects the error to be remedied on remand from the decision of whether there is to be a remand. The important function of the third prong of the plain error test is to prevent a remand for additional proceedings where the defendant, who failed to make a timely objection, cannot show that there is a reasonable probability that a do-over would more likely than not produce a different result. See Dominguez Benitez, 124 S.Ct. at 2340 (explaining that one reason the plain error third-prong burden on the defendant should not be easy is that one of the policies underlying Rule 52(b) is to “reduce wasteful reversals by demanding strenuous exertion to get relief for unpreserved error”). That is why the yardstick of prejudice is change of outcome — a different result if the proceeding were done over with the error corrected. It makes little or no sense to say that there probably would have been a different sentence but for the use of extra-verdict enhancements, and on that basis alone send the case back for another round in which the very same extra-verdict enhancements are to be used again. That is, however, exactly what the Fourth Circuit did in Hughes.
For the third prong of the plain error test to serve its function, the prejudicial impact of the error, which is to say the likelihood of a different result if the error is removed, must be gauged in terms of the part of the procedure that is to be changed. In Booker-error cases that means gauging the likelihood of a different result when the guidelines range is considered advisory instead of mandatory (and any § 3553(a) factors not otherwise taken into account are considered, too). Therefore in order to satisfy the third prong of the plain error test, pr e-Booker defendants must establish a reasonable probability that if the district court had considered the guidelines range it arrived at using extra-verdict enhancements as merely advisory, instead of mandatory, and had taken into account any otherwise unconsidered § 3553 factors, the court would have imposed a lesser sentence than it did. That inquiry makes sense because it hooks the prejudice inquiry to the procedural change necessary to remedy the error.
The only reason the Hughes opinion gives for focusing on the effect of the extra-verdict enhancements, instead of the effect of treating the guidelines range as mandatory, is that the defendant said that was the error committed. This is the Hughes opinion’s explanation:
To demonstrate that the error was prejudicial, Hughes must show that “the error actually affected the outcome of the proceedings.” Here, the error alleged *1303by Hughes was that the district court imposed a sentence greater than the maximum authorized by the facts found by the jury alone. Had the district court imposed a sentence within that maximum, Hughes’ sentence would have been calculated according to an Offense Level of 10 (corresponding to a prescribed range of 6 to 12 months’ imprisonment) — markedly lower than the final Offense Level of 22 (corresponding to a prescribed range of 41 to 51 months’ imprisonment) used by the district court. We therefore conclude that the error affected Hughes’ substantial rights. Accord United States v. Promise, 255 F.3d 150, 160 (4th Cir.2001) (en banc) (holding that Apprendi error resulting in increased sentence affected defendant’s substantial rights).
396 F.3d at 380, 2005 WL 147059, at *5 (footnote and internal citation omitted). In the footnote omitted from that quotation, the Hughes opinion says:
The question for purposes of determining whether Hughes was prejudiced is not what the district court would have done had it imposed a sentence in the exercise of its discretion pursuant to § 3553(a). Hughes does not argue that the district court erred by failing to regard the guidelines as advisory in sentencing him. Rather, Hughes argues that the district court erred by imposing a sentence that was greater than the maximum authorized by the facts found by the jury alone. Therefore, the prejudice inquiry concerns what sentence the court would have imposed had it not committed the error of going beyond the facts found by the jury in imposing a sentence under the mandatory guideline regime then in existence. This case does not present the question of whether a defendant suffers prejudice because a sentencing court fails to treat the guidelines as advisory in determining the sentence.
Id. at 380 n.6, *5 n. 6.
We disagree with the notion that the defendant can define the constitutional error, and thereby predetermine the outcome of the third prong of the plain error test, by the phrasing of his argument. The defendant does not get to define the constitutional error. The Supreme Court defined the constitutional error in Booker, and there the Court said that there would not have been any constitutional error if the guidelines had been used in an advisory instead of mandatory fashion. Booker, 125 S.Ct. at 750 (Stevens, J., maj.op.); id. at 764 (Breyer, J., maj.op.). Defendants cannot reshape that holding more to their liking. The Supreme Court said that the remedy is to use the guidelines in an advisory fashion (along with any § 3553(a) factors not otherwise considered). The prejudice inquiry must focus on what has to be changed to remedy the error, and the consideration of extra-verdict enhancements is not to be changed.
Put differently, regardless of Hughes’ contention to the contrary, the district court did not err by “imposing] a sentence greater than the maximum authorized by the facts found by the jury alone.” Hughes, 396 F.3d at 380, 2005 WL 147059, at *5. The district court erred (understandably) by failing to recognize that it was not required to impose a sentence greater than the maximum authorized by the facts found by the jury alone. It erred in failing to treat the guidelines as advisory. If the court had treated the guidelines as advisory, there would have been no constitutional error.
The correct result of applying the third prong of the plain error test is revealed in the following sentence from the Hughes opinion: “The record does not provide any indication of what sentence the district *1304court would have imposed had it exercised its discretion under § 3553(a), treating the guidelines as merely advisory.” Id. at 381 n. 7, *5 n. 7. As we have explained, where the record does not provide “any indication” that there would have been a different sentence if the court had recognized and exercised its § 3553(a) discretion and treated the guidelines range as merely advisory, the party with the burden of showing a reasonable probability of a different result loses. That party is the defendant.
The second federal court of appeals to address the Booker plain-error issue was the Sixth Circuit. In United States v. Oliver, 397 F.3d 369, 2005 WL 233779 (6th Cir. Feb. 2, 2005), that Court found plain error in a pr e-Booker case where one extra-verdict enhancement increased the sentencing range of the defendant. It reasoned that consideration of the extra-verdict enhancement “unconstitutionally increased Oliver’s sentence beyond that which was supported by the jury verdict and Oliver’s criminal history,” and “[a]s a result Oliver arguably received a sentence that was longer than his sentence would have been absent a Sixth Amendment violation.” Id. at 379-80, *8. That analysis is flawed for several reasons similar to those that affected the Fourth Circuit’s thinking in Hughes.
The Sixth Circuit’s Oliver reasoning assumes that the Sixth Amendment violation is the use of an extra-verdict enhancement. The Sixth Circuit makes the same mistake the Fourth Circuit does by measuring prejudice from use of a factor that the district court may permissibly consider when resentencing on remand. Compare id. with Hughes, 396 F.3d at 380, 2005 WL 147059, at *5. It, like the Fourth Circuit, disconnects the prejudice inquiry from the change to be made.
The Sixth Circuit in Oliver makes another mistake as well. It measures prejudice based on the speculative standard of what is “arguable.” See Oliver, 397 F.3d at 379-80, 2005 WL 233779, at *8 (“As a result Oliver arguably received a sentence that was longer than his sentence would have been absent a Sixth Amendment violation.”) “Arguable” means “susceptible to debate, challenge, or doubt; questionable.” Random House Unabridged Dictionary 112 (2d ed.1993). “Arguable” is not enough to satisfy the third-prong test of prejudice, particularly where it is just as arguable that the pr e-Booker result would not changé in a post-Booker resentencing where the guidelines are only advisory.6
The third federal appeals court to address the Booker plain error issue is the Second Circuit. Its decision in United States v. Crosby, 397 F.3d 103, 2005 WL 240916 (2d Cir. Feb.2, 2005), represents the most novel approach to the Booker third-prong plain error inquiry. The plain eiTor rule is one to be applied by appellate courts in determining whether to vacate or reverse district court judgments, including sentences as well as convictions, based on error that was not preserved by proper and timely objection in the district court. The prejudice inquiry is to be made based on the record before the appellate court. See Dominguez Benitez, 124 S.Ct. at 2340 (Under plain error review, “[a] defendant must ... satisfy the judgment of the re*1305viewing court, informed by the entire record, that the probability of a different result is sufficient to undermine confidence in the outcome of the proceeding.” (emphasis added and internal quotation omitted)); United States v. Vonn, 535 U.S. 55, 59, 122 S.Ct. 1043, 1046, 152 L.Ed.2d 90 (2002) (“[A] silent defendant has the burden to satisfy the plain-error rule and ... a reviewing court may consult the whole record when considering the effect of any error on substantial rights.”). Yet, in Crosby the Second Circuit delegated to the district courts the task of determining whether the appellant has carried his burden of showing a reasonable probability of a different result.
According to Crosby, each Second Circuit pre-Aoofter-error case in which the issue was not preserved by proper objection will be remanded to the district court. On remand the district court is to determine whether it would have imposed a “materially different” or “nontrivially different” sentence under the post -Booker sentencing rules. Crosby, 397 F.3d at 118-19, 2005 WL 240916, at *12. If so, then there is third-prong prejudice, there is plain error, and the district court is to vacate its earlier sentence and resentence the defendant. Id. If the district court decides that it would not have sentenced the defendant to a materially or nontrivially different sentence under the Booker rules, then there is no third-prong prejudice and thus no plain error; the initial sentence remains. Id. The party that loses in the district court is to be allowed to appeal, id. at 119-20, *13, which means the court of appeals will be reviewing the district court’s decision on whether the unpreserved error should have been noticed and corrected the first time it was before the court of appeals.
The Crosby solution turns the review process on its head. The determination of plain error is the duty of courts of appeal, not district courts. See Dominguez Benitez, 124 S.Ct. at 2340 (Under plain error review, “[a] defendant must ... satisfy the judgment of the reviewing court, informed by the entire record, that the probability of a different result is sufficient to undermine confidence in the outcome of the proceeding.” (emphasis added and internal quotation omitted)); Vonn, 535 U.S. at 59, 122 S.Ct. at 1046 (“[A] silent defendant has the burden to satisfy the plain-error rule and ... a reviewing court may consult the whole record when considering the effect of any error on substantial rights.”).
Crosby essentially requires resentencing in order to determine whether resentenc-ing is required. The Second Circuit says that in order for district courts to decide what they would have done had the Booker rules been in effect, the attorneys must be permitted to re-argue the sentencing issues as they would have argued had Booker been the law. Crosby, 397 F.3d at 119—20, 2005 WL 240916, at *13. The district court is then to apply the Booker rules and see what sentence it would have imposed under those rules. There will be a do-over to decide if there should be a do-over. This procedure undermines the Supreme Court’s teaching that one of the principal purposes of the plain error rule is to avoid needless reversals and remands. Dominguez Benitez, 124 S.Ct. at 2340. The Crosby solution will require not only that every sentence in a plain error pre-Booker case be remanded for what amounts to a mock-resentencing, but it will also add the real possibility of another appeal and another remand on top of that.
The Crosby solution cannot be reconciled with the Supreme Court’s quite specific instruction about what federal appeals courts are to do when they cannot tell from the record on appeal whether the unpreserved error affected the outcome of *1306the proceedings. Under Jones, if it is uncertain whether the error affected the result, if the effect is indeterminate, if we simply cannot tell from the appellate record, then the defendant loses. Jones, 527 U.S. at 394-95, 119 S.Ct. at 2105. The Second Circuit admits in Crosby that in pre-Booker cases “we cannot know whether a correct perception of the law would have produced a different sentence.” Crosby, 397 F.3d at 118-19, 2005 WL 240916, at *12. That should have been the end of the inquiry.
The only authority that Crosby cites for turning the plain error process upside down is 18 U.S.C. § 3742(f), which merely provides that: “If the court of appeals determines that — (1) the sentence was imposed in violation of law ..., the court shall remand the case for further sentencing proceedings with such instructions as the court considers appropriate.” Crosby, 397 F.3d at 117, 2005 WL 240916, at *11. Acknowledging that the statutory provision “normally contemplates a remand for resentencing,” Crosby reasons that, because remands for resentencing are authorized, “we think an appellate court necessarily has the lesser power to remand for a determination of whether to resen-tence, and to permit resentencing.” Id. That conclusion does not follow at all. In cases of non-preserved error appellate courts lack the authority to remand for resentencing where the requirements of the plain error rule have not been met. The remands being ordered are not, as Crosby suggests, for the purpose of determining “whether to resentence,” but for the purpose of determining whether the third prong of the plain error test has been met so that the unpreserved error may be noticed and the appealed sentence vacated. Neither § 3742(f) nor any other part of the Sentencing Reform Act purports to give appellate courts the authority to delegate the decision of whether there has been plain error to the very court whose judgment is being reviewed.
We apply the plain error rule at the appellate level, as it is written, and as it has been consistently interpreted by the Supreme Court of the United States. Having done so here, we conclude that Rodriguez has failed to satisfy the third prong of the plain error test by showing that the Booker error in his case affected the outcome of his sentencing within the meaning of Olano, 507 U.S. at 734, 113 S.Ct. at 1777-78, and Dominguez Benitez, 124 S.Ct. at 2340.
AFFIRMED.
. The PSR determined that Rodriguez conspired to distribute 30,000 MDMA tablets, each with an average weight of 147 milligrams, making the total weight of all the tablets 4,410 grams. According to the guidelines drug equivalency table, 1 gram of MDMA is the equivalent of 500 grams of marijuana. U.S.S.G. 2D1.1, cmt. n.10. Thus, 4,410 grams of MDMA is the equivalent of 2,205 kilograms of marijuana.
. Rodriguez attached to his objections a copy of the results of his polygraph examination. These results included that he had answered "yes” when questioned whether (1) his story about being in Tampa on only two occasions was true, and (2) both Salgado and Ruz lied when they testified in court. Moreover, Rodriguez had answered "no” when questioned whether (1) he had come to Tampa on more than two occasions, (2) he met Ruz before his trial, and (3) he had brought more than three packages of MDMA to Tampa for Salgado.
The implication of the polygraph questions, of course, is that Rodriguez had brought three packages of MDMA to Tampa for Salgado, which contradicts Rodriguez's own trial testimony denying any involvement with the drug conspiracy.
. Although the prosecutor initially objected to the admission of the results of this polygraph examination, he subsequently withdrew this objection. The court did admit the results into evidence for whatever they were worth.
. This guidelines range is based on an adjusted offense level of 30 and a criminal history category of I. Because Rodriguez received a minor role reduction pursuant to § 3B 1.2(b), his base offense level was 30, even though the quantity of drugs foúnd by the district court would otherwise have yielded a base offense level of 32. See U.S.S.G. § 2D1.1(a)(3). The two-point enhancement for obstruction of justice and the two-point reduction for minor role canceled each other out, yielding an adjusted base offense level of 30.
. The other enhancement applied was under § 3C1.1 for obstruction of justice because Rodriguez took the stand and falsely denied his guilt. It may be that enhancement was not Booker error because the jury verdict convicting Rodriguez of the crimes he denied *1299necessarily, albeit implicitly, found that he had engaged in behavior that fits within § 3C1.1. We need not decide that question, because the drug quantity was not implicitly found in the jury verdict.
. The Oliver Court also found that the fourth prong of the plain error test was met as a result of the extra-verdict obstruction of justice enhancement applied in that pre-Booker case. 397 F.3d at 379-80, 2005 WL 233779, at *8. Yet, the very next day a different panel of the Sixth Circuit reached exactly the opposite conclusion in another pr e-Booker case involving exactly the same enhancement. United States v. Bruce, 396 F.3d 697, 719, 2005 WL 241254, at *17 (6th Cir. Feb.3, 2005). As we have already mentioned, we need not reach the fourth prong in this case because of our conclusion that Rodriguez has not satisfied the third prong.