WINTER, Circuit Judge.
Samuel Yakobowicz appeals from his conviction by a jury before Judge S. Gwin1 on four counts of filing false federal excise tax returns under 26 U.S.C. § 7206(1) and one count of attempting to impede the administration of internal revenue laws under 26 U.S.C. § 7212(a)(1). These charges arose from excise tax returns filed by Ya-kobowicz on behalf of his company Twenty-Four Hour Fuel Corporation. On appeal, Yakobowicz argues, inter alia, that his right to a fair trial was compromised by a procedure that allowed the government to make summation comments at the conclusion of each witness’s testimony. We agree and vacate the conviction.
BACKGROUND
Appellant owned and operated Twenty-Four Hour, a business that sold and delivered various types of diesel fuel. Among other products, Twenty-Four Hour sold low sulfur and high sulfur diesel fuel. High sulfur diesel fuel, commonly used for home heating, is dyed red and may not be used in on-road vehicles. Low sulfur diesel fuel is clear and may be used in on-road vehicles. Apart from the sulfur content and artificial dye, there is no difference in the composition of the two types of fuel. However, clear low sulfur diesel fuel is subject to a federal excise tax, while the high sulfur product is not. Twenty-Four Hour prepaid the federal excise tax on some of the low sulfur diesel when purchased and was entitled to a refund of prepaid taxes for low sulfur diesel fuel later sold to nontaxable entities. One of Twenty-Four Hour’s customers was the Long Island Railroad (“LIRR”), a tax-exempt entity.
Appellant filed quarterly excise tax returns for Twenty-Four Hour for four quarters in 1996 and 1997 and sought refunds of prepaid taxes on low sulfur diesel that it sold to the LIRR. The IRS disagreed, and Twenty-Four Hour commenced suit in the Eastern District to challenge the IRS’s assessment of tax liability. Twenty Four Hour Fuel Oil Corp. v. United States, 38 F.Supp.2d 217 (E.D.N.Y.1999). Following an extensive hearing, the district court ordered the IRS to reconsider Twenty-Four Hour’s claims for refunds. Id. at 224. This order was later vacated when the IRS agreed to re-examine Twenty-Four Hour’s claims using a new team of tax agents. Twenty Four Hour Fuel Oil Corp. v. United States, 47 F.Supp.2d 1103 (E.D.N.Y.1999).
The IRS’s re-examination led to the present criminal charges against appellant. A five-count superseding indictment charged defendant with four counts of filing false federal excise tax returns in violation of 26 U.S.C. § 7206(1), and one count of corruptly endeavoring to obstruct or impede the administration of the internal revenue laws in violation of 26 U.S.C. § 7212(a).
[147]*147The government’s factual theory was that appellant improperly sought refunds for Twenty-Four Hour by falsely claiming that it had delivered only clear, low sul-phur diesel to the LIRR when in fact some deliveries had been of dyed red home heating oil. The government claimed that appellant submitted false delivery tickets recording sales of dyed red home heating oil to another customer, City Fuel, to account for oil actually delivered to LIRR. The government also claimed that appellant produced and provided his accountants with false information and doctored documents to support his claims for refund.
Before trial, the district judge told the parties that at the conclusion of the testimony of every witness, each party would be allowed to make a short statement to the jury. We style this the “interim summation” procedure. The judge, who appears to have used this procedure rather regularly in the past, explained that “[t]he comments are limited to the statements of that witness and how that testimony fits in the overall scheme of the case.” He further stated that he tried “to make it clear to the jury that [an interim summation] is not testimony and it’s never to be considered as testimony.” Appellant objected on the grounds that the procedure is not authorized by the Federal Rules of Criminal Procedure and gives the government a strong advantage. The judge overruled the objection.
The trial that followed involved mundane factual issues regarding what kind of fuel was delivered to which customers and whether Twenty-Four Hour’s records accurately reflected actual deliveries. The principal witnesses were Twenty-Four Hour’s accountants and truck drivers, as well as IRS agents. In stating how the testimony of particular witnesses “fits in the overall scheme of the case,” the government used the interim summation procedure on several occasions to argue and reargue its theory of the case. Rhetorical questions were posed as to the defendant’s state of mind regarding events recounted by witnesses, and the jury was asked in the interim summations to “decide for yourself’ various critical factual issues. Any distinction between the content of the government’s interim summations and its final summation was all but invisible.
For example, after one Twenty-Four Hour driver, Mr. Koc, testified, the prosecutor stated the following in interim summation:
Mr. Koc is another driver of Twenty Four, who told you that he delivered red dye heating oil to the railroad and recorded it as clear dye diesel fuel.
Now whether the railroad got red dyed oh or clear oil, the fact is he wasn’t delivering what he said he was delivering and Mr. Koc explained that Mr. Yakobowicz explained to him it’s part of his contract with the Long Island Railroad that he had to deliver diesel, so we recorded everything as diesel. That wasn’t what he was delivering. He was delivering something else.
Everyone of the drivers was told to miseharacterize what fuel was going to the Long Island Railroad. That mis-characterization carried over to the records of Twenty Four Hour and it’s carried over to the documents that their accountants would get preparing the tax returns and whether the IRS inspector told him he couldn’t do that, maybe the railroad didn’t care what they were getting, maybe they thought it all burned in the trains, why do we care.
The issue is, what he was doing characterizing the fuel on the tax return? As Ms. Bornstein told you, they weren’t paying the claims for the refunds, but he was not delivering what he said he was delivering to the railroad.
[148]*148We’ve also heard suggestions that Twenty Four Hour, Concord Trucking was delivering product for other companies. But we have Mr. Koc telling you, well, I went to 197 King Street, I put my red dye heating oil into the other trucks.
If he’s delivering product to other companies, why is he loading up these other trucks with the product going to this other companies?
He indicated that he could not remember how many times he had been to 197 King Street. He threw out the numbers five or ten and said that he wasn’t sure. What he was sure of, he never got any delivery tickets to Sea Fuel filled out by anybody else. He never got any deliveries, delivery tickets to Citifuel that he did not fill out himself.
All of the delivery tickets that Mr. Neeman testified that he filled out, what were they used for?
After another driver, Mr. Ozgur, testified, the prosecutor stated:
We have another driver of Twenty Four Hour Fuel, confirms for you, ladies and gentlemen, he was mischaracteriz-ing what he was delivering to the Long Island Railroad. Sam told him to write diesel. Whatever the contract required that’s what should have been delivered.
Mr. Yakobowicz told him to writing [sic] something that wasn’t being delivered there. Yes, he kept a manifest, he kept all of the records, he gave all of it to Mr. Yakobowicz at the end of the day. Mr. Yakobowicz accounted for all of his fuel every day. He knew exactly what product went where. He would know who to charge, how much he was supposed to be getting.
The question is whether Mr. Yako-bowicz accurately accounted for what he was buying and selling to his accountant to produce the tax returns. The manifest reports back to the people in the office what the driver did. The question is whether the people in the office reported to the accountant what was going on.
Mr. Ozgur confirmed he made deliveries to 197 King Street, no more than ten or twenty, might have been less. If you look at the documentation that Twenty Four Hour gave to its accountants, they were selling over a million gallons of oil to Citifuel. Who was driving? You heard from drivers who delivered hundreds of thousands of gallons of oil for Mr. Yakobowicz. It may have been to this 197 King Street and prepared their own particulars.
Mr. Ozgur could not recall whether the tickets were written out for him before they went there, the others drivers did.
After the testimony of IRS Agent Colaiani, the prosecutor made the following statements.
Ladies and gentleman, the testimony of Agent Colaiani is significant in a number of respects. Most important respect are—is it gives you the ability to follow where the fuel went.
Look at his schedules—and Mr. Scaring did a fine job of pointing out he made an error in typing in this information or that information, but look at the schedules themselves. Look at the documents that they corollate to and look at all of the information that he compiled for you, ladies and gentlemen.
What he is doing—what he’s producing for you in these schedules is not the evidence that you need to rely on to make your decision. It is a guide to lead you to the evidence and check the schedules and look at what the evidence shows. Look at what these bills of lading say to pick up. Look at what the bill of—the delivery tickets say the [149]*149trucks should deliver. It doesn’t add up. It doesn’t add up.
And—and Mr. Scaring spent a lot of time talking about how much credit Twenty-Four Hour should get for this or how much credit they shouldn’t get for that.
Well, remember, Mr. Yakobowicz is not charged with tax evasion in this case. The government doesn’t have to prove he owed any tax. In fact, the third quarter 1997, Agent Colaiani testified that the records were so incomplete he can’t figure out how much tax Twenty-Four Hour owed.
The question isn’t whether or not they owed tax. The question is whether or not they accurately recorded the information on their tax return. That’s what Mr. Yakobowicz is charged with, intentionally providing false information on his tax returns.
Decide for yourself whether or not he reported all of the fuel that he bought tax free during each of these quarters. Decide for yourself whether or not he accurately recorded the amount of clear diesel fuel that was going to the Long Island Railroad or whether or not he picked up heating oil, said it was clear diesel fuel and gave records to his accountant saying it was all clean diesel fuel.
Decide for yourself whether or not he provided accurate information to his accountants. Decide for yourself whether or not he accurately subscribed to those tax returns.
He also tells you about this audit that was conducted in 1999. He described to you the Citifuel documents that he received during the course of that audit. Those Citifuel documents were to be used for him to analyze and try to figure out where the fuel went.
The Twenty-Four Hour didn’t provide him the records in a format that made it easy for him to figure out where the fuel went. Decide for yourselves why wouldn’t they want to make it easy for the IRS to figure out where the fuel went if they wanted their refund they were being audited for.
The CitiFuel documents were used to explain home eat [sic] heating fuel that was going to the Long Island Railroad and the summary schedule that he produced of those CitiFuel documents illustrates clearly to you that these documents were phoney documents. These documents were double punched delivery tickets, double punched when deliveries were being made to the Long Island Railroad, just like the double punch document that we showed you right there, that was left in the files of Twenty-Four Hour.
We don’t have the CitiFuel records anymore. We do have the double punch ticket.
The defense response by way of interim summation was largely along the lines of pointing out the lack of concealment by appellant. The defense presented no witnesses. Appellant was convicted on all five counts. He now appeals on the ground, inter alia, that the interim summation procedure denied him a fair trial.
DISCUSSION
The government argues that we should review the district court’s decision to employ interim summations for abuse of discretion out of “reluctan[ce] to interfere with district judges’ management of their very busy dockets.” Whiting v. Lacara, 187 F.3d 317, 320 (2d Cir.1999) (reviewing denial of attorney’s motion to withdraw for abuse of discretion); United States v. Blackwood, 456 F.2d 526, 529-30 (2d Cir.1972) (district court’s refusal to allow re[150]*150call of witness reviewed for abuse of discretion). We agree that a trial management issue is involved and review for abuse of discretion.
We first note that the procedure used here allowed fully argumentative summations. The prosecution not only stated what the particular witness (and sometimes prior witnesses) said but also argued the inferences to be drawn by the jury. Rhetorical questions were repeatedly posed as to appellant’s state of mind regarding events recounted by witnesses, and the jury was asked to “decide for yourself’ various critical issues.
The standard practice in criminal cases is to allow summations only after the closing of the evidence and just before the trial judge’s instructions to the jury. This is the clear anticipation of Fed.R.Crim.P. 29.1, which states that the closing argument of the government be first, then that of the defense, followed by the government rebuttal. Fed.R.Crim.P. 29.1 (2002). This ordering of summations was made effective in 1975. It reflected the view that the “fair and effective administration of justice is best served if the defendant knows the arguments actually made by the prosecution in behalf of conviction before the defendant is faced with the decision whether to reply and what to reply.” Fed.R.Crim.P. 29.1 advisory committee’s note B. to 1975 Enactment. This version of Rule 29.1 also stated that the closing arguments were to be held “[ajfter the closing of evidence”. Fed.R.Crim.P. 29.1 (1975). Although the current version omits that language, the Advisory Committee Notes make clear that the amendment eliminating the language was part of an overall restyling of the criminal rules that was to be “stylistic only.” Fed.R.Crim.P. 29.1 advisory committee’s notes to 2002 Amendments.
The traditional order of events at a criminal trial'—opening statements, presentation of evidence, summations, and jury instructions—has numerous purposes. Among these purposes is to enable juries to avoid forming opinions before the close of evidence and deliberations. This is reflected in trial courts’ repeated instructions to juries to keep an open mind until deliberations. The traditional order also aids the defense in relying upon the presumption of innocence as a principal defense by forcing the prosecution to set forth its entire theory before a defense is framed.
Those goals require that a distinction between evidence and argument be observed until summations. Opening statements are generally confined to expectations as to what the evidence will show and usually do not contain large amounts of argument. See United States v. Dinitz, 424 U.S. 600, 612, 96 S.Ct. 1075, 47 L.Ed.2d 267 (1976) (Burger, C.J., concurring) (“An opening statement has a narrow purpose and scope. It is to state what evidence will be presented, to make it easier for the jurors to understand what is to follow, and to relate parts of the evidence and testimony to the whole; it is not an occasion for argument.”) (emphasis supplied); Thomas A. Mauet & Warren D. Wolfson, Trial Evidence 29-43 (3d ed.2005) (explaining that opening statements should not mention inadmissible or unprovable evidence, should not be argumentative, and should not state the attorney’s personal opinions or mention the opponent’s case); see also United States v. Zielie, 734 F.2d 1447, 1455 (11th Cir.1984) (trial judge can exclude irrelevant facts from opening statements and stop arguments from being made in them); United States v. Salovitz, 701 F.2d 17, 21 (2d Cir.1983) (recognizing that opening statements are not occasions for argument). [151]*151During the presentation of evidence one of the most commonly sustained objections is that a particular question is argumentative, Fed.R.Evid. 611(a) advisory committee’s note to Subdivision (a) to 1972 Proposed Rules, and any summation-like remarks by counsel during the presentation of evidence are improper and subject as a routine matter to being stricken, Mauet & Wolfson, supra, at 30.
However, interim summations in some form have been permitted in lengthy and/or complex civil trials. See In re Brooklyn Navy Yard Asbestos Litigation, 971 F.2d 831, 836 (2d Cir.1992) (as part of handling the consolidated trial of 79 asbestos-related personal injury and wrongful death actions “to ensure that the jurors could assimilate the vast amounts of information necessary to assess the claims,” the district court employed an interim summation procedure); In re New York Asbestos Litigation, 149 F.R.D. 490, 499 (S.D.N.Y.1993) (interim summations “considered and adopted where appropriate during the consolidated trial” of tort actions based on asbestos exposure). Use of such procedures in complex and lengthy civil trials has been looked upon favorably by commentators. See e.g., Tom M. Dees, III, Juries: On the Verge of Extinction? A Discussion of Jury Reform, 54 S.M.U. L.Rev. 1755, 1778-80 (2001) (summarizing arguments for and against interim summations and citing state task forces advocating use of said); Douglas G. Smith, Structural and Functional Aspects of the Jury: Comparative Analysis and Proposals for Reform, 48 Ala. L.Rev. 441, 537 (1997) (interim summations desirable in lengthy or complex cases). There does not, however, appear to exist any authority or advocacy for the use of interim summations in criminal cases, much less in the form used here.
We believe that use of this procedure in the present matter is not justified by the authorities cited. The district court assumed that use of this or similar procedures is not prohibited in every criminal case—a point not before us and not decided—and that use in criminal eases did not raise issues not generally present in civil cases—a point decidedly before us and discussed infra. Even if those assumptions were correct, use of the procedure should be based on findings that the case at hand differs from the garden variety of cases in which summations only at the close of evidence are sufficient. These differences exist, if at all, in the length of a trial or the complexity of the issues, and most commonly in a combination of the two. The present case involved no length, no complexity, and no need.
The presentation of evidence, including opening statements, began on a Monday afternoon and ended on the next Friday, four and one-half days later. The final summations and jury instructions took place on Monday; the verdict was reached on Tuesday. The key witnesses were Twenty-Four Hour’s accountants, its truck drivers who delivered fuel and filled out delivery tickets, and two IRS Agents who had performed an audit of Twenty-Four Hour. The issues were not entirely simple but were certainly not complex, as is reflected in the interim summations quoted above. The issues largely involved what kind of oil was delivered to certain customers, who prepared particular invoices, and whether some invoices accurately reflected deliveries. There was, therefore, no demonstrated need for the departure from standard practice in allowing interim summations. In fact, the district court did not view this case or the interim summation procedure as exceptional. If we allow use of the procedure here, therefore, we will have to allow it in virtually all criminal cases.
[152]*152The use of interim summations raises very different issues in civil and criminal cases. Bifurcation of issues is not uncommon in civil cases, and special verdicts separating the disposition of discrete factual and/or legal issues are specifically authorized by the Federal Rules of Civil Procedure. See Fed.R.Civ.P. 42 (permitting bifurcation “of any separate issue or any number of claims”); Fed.R.Civ.P. 49(a); United States Football League v. National Football League, 842 F.2d 1335, 1365-66 (2d Cir.1988) (reproducing entire special verdict form in complicated antitrust case). Multiple summations in civil cases are not, therefore, inconsistent with the judicial system’s growing allowance of separate dispositions of discrete issues in civil cases.
Also, in civil cases, the procedure does not systematically favor plaintiffs or defendants. Even if uncontrolled, the procedure is likely to have random effects as between plaintiffs and defendants. Moreover, control of interim summations is easier in civil than in criminal cases. In a civil case, control of what is properly said and when in an interim summation can be calibrated to discrete issues of the particular case in order to foreclose an undue advantage to any party. This calibration—limiting interior summations to certain points in the trial, restricting the scope of argument, etc.—is facilitated by pretrial discovery that allows the parties to see or hear all the evidence ahead of time. A pretrial conference and ensuing pretrial order allows the appropriate controls to be in place largely before the first witness is called.
By way of contrast, in criminal cases there is no bifurcation, at least as to each count of an indictment, as to the guilt determination, and general verdicts are strongly preferred. United States v. Pforzheimer, 826 F.2d 200, 205 (2d Cir.1987) (“[JJury interrogatories in criminal cases are generally disfavored in this Circuit ....”).
In addition, the interim summation procedure will, in the great bulk of criminal cases, systematically strengthen the prosecution’s case. This is obvious even at the most superficial level. In criminal cases, the prosecution almost always calls more witnesses than the defense, which not uncommonly calls none. Interim summations after each witness enable the prosecution to argue repeatedly the merits of its theory of the case. Because a party’s witness generally gives testimony favoring that party, the proponent of a witness has more to work with in its interim summation than does the adversary. The effect of interim summations in most criminal cases is, therefore, to strengthen the government’s theory cumulatively as well as repetitively.
At a somewhat deeper level, the advantage grows. There is limited discovery by defendants in criminal cases, whereas the prosecution has grand jury subpoenas at its disposal. The government generally has, therefore, a clearer vision of the entire case than does the defense and can unveil its evidence with interim summations in mind. Given this informational disadvantage, the defense may find it very risky to respond to particular interim summations by emphasizing evidentiary gaps that may be filled immediately thereafter or by promising or implying a defense that is ultimately not presented. A failure to respond to the government’s interim summation, on the other hand, leaves the government with a growing advantage.
Finally, a criminal defendant is presumed innocent. Unless the prosecution proves guilt beyond a reasonable doubt, the defendant is entitled to acquittal without having to offer evidence “(or indeed lift a finger) in his defense.” United States v. Jackson, 368 F.3d 59, 66 (2d Cir.2004). The Supreme Court has gone [153]*153so far as to describe the presumption of innocence as a “shorthand description” of the criminal defendant’s right “to remain inactive and secure, until the prosecution has taken up its burden and produced evidence and effected persuasion .... ” Taylor v. Kentucky, 436 U.S. 478, 483 fn. 12, 98 S.Ct. 1930, 56 L.Ed.2d 468 (1978) (internal quotation marks and citations omitted).
We conclude that use of the interim summation procedure in the present case was not only unjustified by the authorities cited above but also violated appellant’s constitutional right to a fair trial. For better or for worse, our system of justice anticipates that a criminal defendant is entitled to see the prosecution’s whole case before deciding on a defense and to be judged by a jury that is strongly warned to keep an open mind until deliberations. This scheme seeks to reduce the tactical disadvantage of a defendant’s waiving an opening statement or limiting such a statement to cautioning the jury to withhold judgment until the end. It seeks to allow a defendant not to call any witnesses and still argue for an acquittal. It is a process that increasingly locks in the prosecution to a particular theory of the case while putting off the defense’s commitment to a theory until the last practical moment and even then allowing the defense merely to seek to poke holes in the government’s case. Important to this scheme is Rule 29.1’s stricture that the defendant’s summation follows the prosecution’s.
The use of interim summations in the present case materially altered this scheme. The government was allowed to repeat and reinforce with advocacy the testimony of its witnesses. Rhetorical questions as to the conduct and intent of the defendant were posed by the prosecution at several points in the presentation of its main case. At each of these points, the defense was forced to choose either to respond or appear to be acquiescing in the government’s argument. Instead of being allowed to assert the presumption of innocence and poke holes after viewing the prosecution’s case as a whole, the defense had, as a practical matter, to respond after the testimony of several witnesses. Moreover, the jury was asked again and again by the prosecution to form opinions—“decide for yourself’—on key issues before the government rested. To be sure, the judge repeatedly told the jury not to form such opinions, but allowing interim summations was so inconsistent with those cautions that jury confusion was quite likely-
A constitutional error does not always require reversal of a criminal conviction. Chapman v. California, 386 U.S. 18, 22, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967). There are two kinds of errors: “ ‘trial errors’ which are of relatively limited scope and which are subject to harmless error review, and ‘structural defects,’ which require reversal of an appealed conviction because they ‘affect [] the framework within which the trial proceeds.’ ” United States v. Feliciano, 223 F.3d 102, 111 (2d Cir.2000)(error in conducting portion of voir dire outside defendants’ hearing not structural error requiring reversal) (quoting Arizona v. Fulminante, 499 U.S. 279, 307-10, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991) (holding harmless error rule applies to admission of involuntary confessions)) (alterations in original). “Errors are properly categorized as structural only if they so fundamentally undermine the fairness or the validity of the trial that they require voiding its result regardless of identifiable prejudice.” Yarborough v. Keane, 101 F.3d 894, 897 (2d Cir.1996). Some examples of structural errors include deprivation of the right to counsel; trial by a biased judge; “unlawful exclusion of [154]*154members of the defendant’s race from a grand jury;” preventing defendant from representing himself at trial; and refusal to afford defendant a public trial. Arizona v. Fulminante, 499 U.S. 279, 309-310, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991). Conversely, errors that do not affect the framework of a trial, but rather are discrete events that occur during the presentation of the case and may be “quantitatively assessed in the context of the other evidence presented in order to determine whether its admission was harmless beyond a reasonable doubt,” do not automatically require reversal. Id. at 307-08.
Allowing argumentative interim summations in this case was a structural error requiring reversal. The “entire conduct of the trial from beginning to end is obviously affected,” id. at 309-10, 111 S.Ct. 1246, by a procedure that systematically allowed argumentative summations after each witness without any showing of particularized need based on length of the trial or complexity of the issues, without any authorization in a rule of criminal procedure, and without any attempt to limit the argumentative aspects of the interim summations.
The problem is not that any particular interim summation was unduly prejudicial. The problem is that the repetitive and cumulative summations altered and undermined the defense’s use of the presumption of innocence as a defense and had indeterminable effects on defense strategy and tactics. It is simply beyond the power of harmless error analysis to determine the impact of a procedure that had a repetitive and cumulative effect. It may be that the government’s case would have been overwhelming even using the traditional structure. All we can discern at this stage, however, is that the interim summations certainly made it seem overwhelming.
The interim summation procedure was, therefore, a “structural defect affecting the framework within which the trial proceeded], rather than simply an error in the trial process itself.” Id. at 310. Imposition of this procedure systematically deprived defendant of his right to a fair trial and constitutes structural error requiring reversal.
Having said that, we also note that the gravamen of our concerns has to do with argumentative summations, at least as presented in the context of this case. Nothing we say is intended to prevent the use of procedures that aid jurors in lengthy and/or complex cases by clarifying what elements of a criminal case are the subject of a witness’s testimony or by clarifying what issues are then at stake as the presentation of the evidentiary portion of the case proceeds. The model for such procedures, if a showing of a need for them is made, is to be found in opening statements which, as noted, are generally limited to statements of expectations as to the evidence rather than arguments.
CONCLUSION
For the reasons stated above, the judgment of the district court is vacated.