Responding to growing concerns about widespread abuse of OxyContin, a Schedule II narcotic opioid often prescribed to treat chronic pain, the federal Drug Enforcement Administration (“DEA”) in 2001 launched a public campaign called the “Ox-yContin Action Plan” to ferret out unlawful uses of the drug. Dr. David Demaret Chube II and his brother Dr. Charles Randall Chube (“Dr. David” and “Dr. Randy,” respectively, or, collectively, “the Doctors”) were two of the hundreds of physicians investigated by the DEA for possible illegitimate prescribing of the drug. On February 2, 2005, the Doctors were charged in a 33-count indictment with unlawful distribution of controlled substances, health care fraud, and conspiracy to commit each of those offenses. After a two-week jury trial, the jury acquitted Dr. Randy of 32 out of 33 charges, and acquitted Dr. David of 27 out of 33 charges, rejecting both the conspiracy charges and many distribution charges. It found Dr. Randy guilty of one count of unlawful distribution and Dr. David guilty of four *695counts of unlawful distribution and two counts of defrauding a health benefit program.
After the sentencing hearing, at which relevant conduct findings played a critical role in enhancing each brother’s advisory Guidelines range, the district court sentenced Dr. Randy to five years’ imprisonment and Dr. David to 15 years. Both men appeal. We affirm their convictions, but we vacate both sentences and remand for resentencing.
I
The Doctors jointly owned a clinic, Great Lakes Family Health Center, which opened its doors in 1998 in Gary, Indiana; they opened a second office two years later in nearby Munster, Indiana. Prior to starting the Great Lakes clinic, the two had practiced medicine with their father in Gary. During the years that the Doctors operated their clinics, many patients came to them seeking relief from severe chronic pain. Like many practitioners, the Doctors treated some of these complaints with OxyContin, a drug that has received praise from pain-management organizations and specialists for its ability to alleviate debilitating pain. From 1995 to 2001, the drug’s maker, Purdue Pharma, openly (and, we now know, falsely) marketed OxyContin to physicians as a less-addictive alternative to other pain-relieving drugs. Because of an emerging realization that OxyContin was addictive and thus prone to abuse, the drug eventually attracted the DEA’s attention.
The DEA was led to the Doctors by one of their patients, William Perry Mitchell, who lived in Benton Harbor, Michigan, about 70 miles from the Great Lakes clinic in Gary. He was one of several patients from that area. Although the Doctors had several legitimate patients, the proof at trial showed that others had no real medical complaints and went to the Doctors’ clinic solely to obtain OxyContin. Mitchell was arrested on September 17, 2001, and charged in the U.S. District Court for the Western District of Michigan with knowingly and intentionally distributing Oxy-Contin pills. Mitchell and his girlfriend had obtained the pills in question using prescriptions written by either Dr. David or Dr. Randy. In exchange for a provision in his plea agreement offering a possible reduction in his sentence, Mitchell agreed to name his “suppliers,” to testify against them, and to bring more witnesses to the Government who would do the same. Mitchell fulfilled all parts of his bargain, as did the Government.
The parties’ briefs present starkly different portraits of the defendants and their conduct. (We note, however, that at this stage we must view the facts in the light most favorable to the jury’s verdict. See Jackson v. Virginia, 443 U.S. 307, 319, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979); United States v. Thompson, 523 F.3d 806, 809-10 (7th Cir.2008).) According to the Doctors, the evidence demonstrated that they believed, in good faith and with good reason, that their patients were seeking treatment for true medical complaints. The problem they face is that the jury did not have to accept their protestations. But the Doctors also raise a legal argument: their convictions, they argue, assess their actions by reference to the standard of care applicable in a civil malpractice suit, but the proper standard is the one found in the Controlled Substances Act (“CSA”), which authorizes the conviction of a registered practitioner only if the prescription was written without a legitimate medical purpose and outside the scope of professional practice. The Government urges us to conclude that the evidence supports a finding that the Doctors were not using their medical licenses to treat *696patients with true complaints, but were acting as common drug dealers, earning substantial amounts of money by prescribing drugs to addicts whom they knew had no legitimate medical complaints and without conducting sufficient physical examinations, diagnostic tests, or other appropriate inquiries or procedures to determine that the prescriptions were warranted. The jury found, the Government continues, that this conduct violated the CSA and thus went beyond simple malpractice. In other words, it found that the Doctors were acting not as physicians, but as profiteering pill-pushers.
The jury drew careful lines in its verdict. It exonerated the Doctors on the great majority of the charges, but it did convict Dr. Randy on one count of unlawful distribution, and Dr. David on four counts of unlawful distribution and two counts of health care fraud. On appeal, the Doctors support their argument about the erroneous use of the malpractice standard with an attack on two of the Government’s expert witnesses, Dr. Theodore Parran and Dr. Robert Barkin. Their testimony allegedly conflated the civil and criminal standards of care and thus created a risk that the jury found liability not because it concluded that the Doctors’ acts of prescribing medications fell outside the scope of legitimate medical practice, but instead because it thought they had been careless. The Doctors also argue that the experts’ testimony should not have been admitted because each impermissibly testified to legal conclusions.
The Government’s case was not limited to these two experts. The jury also heard from 15 patients, 11 of whom testified that they were suffering from true medical problems when they consulted the Doctors; the other four confessed that they fabricated their complaints solely to obtain painkillers. All said that they reported severe pain to the Doctors. Those who fabricated their complaints said they did not tell the Doctors that they were lying or that they were addicted to the drugs, for doing so would have thwarted their efforts to obtain the pills. The battleground of the litigation, then, was whether the Doctors knew that no legitimate medical reason existed for prescribing painkillers to these patients.
At sentencing, the district court’s relevant conduct findings dramatically enhanced each defendant’s advisory Guidelines range. Before adding the relevant conduct, Dr. Randy was facing an advisory Guidelines range of 0 to 6 months’ imprisonment; because this fell within Zone A of the Guidelines grid, probation alone would have been permissible. Dr. David was looking at an advisory Guidelines range of 21 to 27 months in prison. Relying primarily on spreadsheets of alleged unlawful prescriptions compiled by the Government, which included any prescription for a controlled substance found in any of the 98 patient files seized in the Government’s searches of the defendants’ clinics, and on the expert testimony of Dr. Parran, the district court found that each defendant was responsible for all controlled substances — including (among others) Oxy-Contin, Vicodin, and Xanax — that either doctor had prescribed to the patients whose charts had been admitted into evidence at trial. The court then sentenced Dr. Randy to 60 months’ imprisonment, and Dr. David to 180 months. We discuss below additional details of the sentencing proceedings, where relevant.
II
We first address the arguments that the Doctors raise against their convictions. They focus on the expert testimony of two Government witnesses, Dr. Theodore Par-ran and Dr. Robert Barkin. Dr. Parran, *697who specializes in internal medicine and addiction medicine, evaluated all 98 patient files in the record. Based on that review, he concluded that the prescribing “was not done consistent with the usual standards of medical practice” and thus was not done with a “legitimate medical purpose.” Dr. Barkin was called as an expert on pharmacology. Though not a medical doctor, Dr. Barkin received his doctorate in clinical pharmacy in 1985 and is board-certified by various associations for pain management and forensic medicine. Like Dr. Parran, Dr. Barkin testified solely on the basis of the patient charts, although he reviewed only a selection. He, too, concluded that the prescriptions in the charts that he reviewed were issued “[ojutside the scope of medical practice, not for legitimate purposes.”
The Doctors offer two reasons why both experts’ testimony should have been excluded in response to their motion in limine. They have an uphill battle, because our review is only for abuse of discretion. United States v. Watts, 95 F.3d 617, 619 (7th Cir.1996). The scope of their motion was the subject of some dispute. The Government describes it as a “breathtakingly broad” motion that sought to rule out all expert testimony that would suggest a violation of the standard of care applicable in civil medical malpractice cases. This, the Government argued, went too far. While it conceded that the expert testimony would not be conclusive on the question of the Doctors’ criminal liability, it argued that “such evidence was relevant to circumstantially establishing that the defendants had knowingly and intentionally distributed drugs as mere pill-pushers rather than in the course of a professional medical practice.” For their part, the Doctors protest that they have at all times recognized that the experts’ testimony had some relevance. The goal of their motion was only somehow to limit the admissibility of such evidence when it tended to conflate the civil and criminal standards, not to exclude it entirely. But the memorandum supporting the defendants’ motion offers more support for the Government’s position:
[t]he purpose of this Motion in Limine is to request that this Court enter a preliminary ruling prohibiting the Government from introducing any evidence at trial that the Chubes’ treatment of patients did not conform to the “standards of medical practice”, or any other evidence that would be suggestive of a violation of the civil standard of care applicable in medical malpractice cases.
The district court was entitled to take the Doctors at their word. On that understanding it did not abuse its discretion in denying the motion in limine.
The Doctors also argue that the district court committed reversible error when it failed to exclude or strike the evidence during the trial, once it became clear that the testimony was creating precisely the type of confusion that the motion in limine sought to prevent. The net result, they assert, was effectively to reduce the Government’s burden from the standard of criminal intent to the negligence requirement that applies to civil malpractice. Furthermore, they argue, the experts provided what amounted to impermissible legal conclusions on the ultimate question of the Doctors’ intent. We address these two points in turn.
A
In order to support a violation of the CSA, the jury had to find that the Doctors knowingly and intentionally acted “outside the course of professional practice” and without “a legitimate medical purpose.” An implementing regulation issued under the CSA, 21 C.F.R. § 1306.04, reiterates *698this standard: “A prescription for a controlled substance[,] to be effective[,] must be issued for a legitimate medical purpose by an individual practitioner acting in the usual course of his professional practice.” See, e.g., United States v. Bek, 493 F.3d 790, 798 (7th Cir.2007) (“[T]o convict ... a practitioner registered to distribute controlled substances[ ] of violating § 841(a)(1), the government must show that he prescribed controlled substances outside ‘the course of professional practice.’ ”); see also United States v. Moore, 423 U.S. 122, 138-43, 96 S.Ct. 335, 46 L.Ed.2d 333 (1975). As one court summarized it:
[T]o convict a practitioner under § 841(a), the government must prove (1) that the practitioner distributed controlled substances, (2) that the distribution of those controlled substances was outside the usual course of professional practice and without a legitimate medical purpose, and (3) that the practitioner acted with intent to distribute the drugs and with intent to distribute them outside the course of professional practice. In other words, the jury must make a finding of intent not merely with respect to distribution, but also with respect to the doctor’s intent to act as a pusher rather than a medical professional.
United States v. Feingold, 454 F.3d 1001, 1008 (9th Cir.2006) (emphasis in original).
When all is said and done, we agree with the Government that it is impossible sensibly to discuss the question whether a physician was acting outside the usual course of professional practice and without a legitimate medical purpose without mentioning the usual standard of care. It is true that the experts did not, every time, spell out the fact that something more than conduct below the usual standard of care was needed to show an absence of a valid medical purpose. Even the district court was not always as clear as it might have been (although as far as we can tell it never misspoke within the hearing of the jury). During a pretrial motions hearing, for example, the district court indicated its belief that reliance on the civil standard of care could be a permissible theory of the case for the Government:
[B]oth sides are entitled to put in their theory of the case. And if [the prosecution’s] theory of the case is that these doctors have dispensed drugs improperly because they didn’t do the proper work-up, that may be a question of fact for the jury. You may not like the way they do it, but I don’t know that the government can’t do that.
The following exchange at the trial, however, was more typical:
Q. [by the prosecution]: Doctor, would you like me to repeat the question?
THE WITNESS: I believe I recall it pretty well.... It is never appropriate to write a prescription for the spouse of a patient when that prescription is intended for the patient; even more so when it’s a Schedule II narcotic.... It’s not consistent with the usual course of medical practice.
Q. And that would not be for a legitimate medical purpose, correct?
A. Correct.
THE COURT: Counsel, this is being asked regarding standard of care, not legality?
[PROSECUTION]: Absolutely, your Honor.
Thus, what the jury heard was (1) an opinion from the expert that no legitimate medical purpose existed for the prescription in question; and (2) a clarification from the court that the “standard of care” is an issue distinct from the question of “legality.” The former was just what de*699fense counsel, during a sidebar immediately preceding this exchange, had argued that Dr. Parran could testify to, and the latter reflected the distinction that the Doctors now claim was not properly drawn during the trial. We are satisfied that the district court did not abuse its discretion in permitting this line of questioning and that a properly instructed jury could keep the relevant concepts straight. Given this finding, we need not address the Doctors’ failure at crucial points to object to this line of inquiry.
The Doctors also argue that the district court’s charge to the jury was insufficient to cure the confusion created by the experts’ testimony. Given the practical reading we give to jury instructions, see United States v. Matthews, 505 F.3d 698, 704 (7th Cir.2007), we find no merit in this point either. The district court’s instructions to the jury contained no inaccurate statements of the law. Viewing the charge as a whole, we see several points at which the instructions make clear that unlawful-distribution liability cannot attach unless no legitimate medical purpose existed for the prescription. The charge elaborated on the meaning of the phrases “in the course of professional practice” and “no legitimate medical purpose”:
A controlled substance is prescribed by a physician in the course of his professional practice, and therefore lawfully, if the substance is prescribed by him in good faith in medically treating a patient.
Good faith means good intentions and the honest exercise of good professional judgment as to a patient’s medical needs. Good faith means an observance of conduct in accordance with what the physician should reasonably believe to be proper medical practice.
In order to determine whether or not a prescription or prescriptions were issued in the course of a defendant physician’s professional practice, you may consider all of the evidence of circumstances surrounding the prescribing of the substance in question, the statements of the parties to the prescription transactions, any expert testimony as to what is the usual course of medical practice, and any other competent evidence bearing on the purpose for which the substances in question were prescribed.
Unless you find beyond a reasonable doubt that an act of prescribing charged in the Superceding Indictment was not done in the course of his professional practice, then you should find the defendant you are considering not guilty of the charge you are considering.
In addition, the court permitted defense counsel to draw out the distinctions between the civil and criminal burdens during opening statements, cross-examinations, and closing arguments.
Though it is true that the jury instructions did not spell out the distinction between the civil and criminal burdens of proof as expressly as the court did in a case reviewed by the Fourth Circuit, see United States v. Alerre, 430 F.3d 681, 687 & n. 5 (4th Cir.2005), there is no one right way to convey the governing standards. This is particularly true where, as here, the defense made no effort even to propose the desired instruction. If it were vital to the defense that the jury receive further clarification on this issue, then the defense should have submitted a proposed instruction. In sum, the district court did not abuse its discretion in allowing this expert testimony. Its instructions to the jury accurately described the governing standards, and the Doctors’ failure to make any contemporaneous objection or to propose an alternative or additional instruction was fatal to their claim on appeal for reversible error.
*700B
The Doctors’ second challenge to the Government’s expert witnesses is that their testimony invaded the province of the jury by giving opinions on the ultimate legal question whether they knowingly violated the law. After raising this objection in their motion in limine, the Doctors did not repeat it during the trial. If the ruling on the motion in limine was “definitive,” then this was enough to preserve the argument. See Fed.R.Evid. 103(a). Here, the court did not signal any willingness to reconsider its ruling during the trial, and so we apply the usual abuse of discretion standard to this part of the case.
The question whether the district court improperly allowed the prosecution’s experts to testify as to impermissible legal conclusions boils down to an inquiry into the court’s application of FedH.Evid. 704, which permits an expert to testify about an “ultimate issue to be decided by the trier of fact,” Rule 704(a), but nonetheless prohibits the expert from stating “an opinion or inference about whether the defendant did or did not have the mental state or condition constituting an element of the crime charged or of a defense thereto,” Rule 704(b). We must decide whether, in opining that the Doctors wrote prescriptions with no legitimate medical purpose, the experts in this case crossed the line established by Rule 704(b).
The Government argues that neither Dr. Parran nor Dr. Barkin ever claimed to know the Doctors’ intent, and so the opinions that they offered were not barred by Rule 704(b). Particularly since the defense raised no contemporaneous objection, we agree with this position, though we note that portions of Dr. Parran’s testimony come close to a statement about the Doctors’ mental state. For example, when testifying about various “red flags” that signal drug-seeking behavior, Dr. Parran stated that these flags result in a situation where a doctor is “knowingly,” rather than “inadvertently,” “doing harm to a patient.” He then said that when enough red flags have appeared, one can say that a doctor “knew or should have known that harm was being done with these prescriptions.” But these statements are phrased in general terms (“a” doctor in “x” situation) and do not refer directly to the defendants. The district court thus did not abuse its discretion when it did not, on its own motion, intervene and halt this line of inquiry.
The Doctors’ case is strikingly similar to one from the Eighth Circuit, United States v. Katz, 445 F.3d 1023 (8th Cir.2006). Notably, the expert whose testimony was at issue in Katz was none other than Dr. Theodore Parran. In Katz, the Eighth Circuit rejected the defendant’s argument that Dr. Parran had impermissibly testified about Dr. Katz’s criminal intent. Acknowledging that “Rule 704(b) ‘prohibits experts from stating an opinion as to whether the defendant had the requisite mental state for the crime charged,’ ” the court was nevertheless satisfied that “Dr. Parran did not testify regarding the subjective mental state of Dr. Katz upon writing the prescriptions charged in the indictment.” Id. at 1032. The same is true in this case, for while Dr. Parran stated his opinion that no legitimate medical purpose justified the prescriptions in the files he reviewed, he repeatedly cautioned that he was looking only at the files and that he had never had any contact with either the patients or the Doctors. The same applies to Dr. Barkin’s testimony.
The defense urges us to reject Katz, but we see no reason to do so. Its reasoning is sound, and it is consistent with similar decisions from this court. In United States v. Glover, 479 F.3d 511 (7th Cir.2007), for example, we confronted a chai-*701lenge to testimony from a law enforcement official that stated an opinion about the criminal nature of a defendant’s activities. We noted that this court has held on multiple occasions that “ ‘such testimony should not be excluded under Rule 704(b) as long as it is made clear, either by the court expressly or in the nature of the examination, that the opinion is based on the expert’s knowledge of common criminal practices, and not on some special knowledge of defendant’s mental processes.’ ” Id. at 516 (quoting United States v. Lipscomb, 14 F.3d 1236, 1242 (7th Cir.1994)). We also have upheld the admission of expert testimony “to the effect that financial transactions did not comply with regulations and appeared to be fraudulent.” United States v. Davis, 471 F.3d 783, 789 (7th Cir.2006) (citing United States v. Owens, 301 F.3d 521, 526-27 (7th Cir.2002)).
We conclude that the district court did not abuse its discretion in admitting the testimony of Drs. Parran and Barkin. The court repeatedly told the jury that only the court can instruct the jury on what is or is not legal, and the expert witnesses in this case did not go so far as to offer an opinion on the Doctors’ subjective intent.
Ill
We turn now to the Doctors’ challenge to their sentences. They raise a single, though central, argument: that the district court erred in its determination of relevant conduct for purposes of U.S. Sentencing Guideline (“U.S.S.G.”) § 1B1.3. At the sentencing hearing, the district court concluded that although the jury had acquitted the defendants on most of the charged offenses, the Government had established by “clear and convincing evidence” that the defendants were guilty of all of the conduct described in the Presen-tence Investigation Reports (“PSRs”). The district court noted that the less rigorous preponderance-of-the-evidence standard likely would have been sufficient, but in light of lingering doubt on this question at the time of the hearing, the court opted for the higher clear-and-convincing threshold. Since then, United States v. Reuter, 463 F.3d 792 (7th Cir.2006), has made it clear that it is the preponderance standard that applies to findings by a judge during sentencing. But even when we apply the proper preponderance standard to this record, we cannot conclude that the Government’s evidence was sufficient to include as “relevant conduct” all of the activities described in the PSRs.
A
We begin by emphasizing that the burden in the sentencing proceedings was on the Government to show that a given prescription had no legitimate medical purpose and was not dispensed in the usual course of medical practice. To meet that burden, the Government produced spreadsheets that listed every prescription for a controlled substance contained in the 98 patient files in evidence. The spreadsheets listed the patient’s name, the date of the prescription, the type of pill prescribed, and the dose. The probation officer used the spreadsheets and the pill totals that they tallied for the PSRs’ relevant conduct calculations. Those totals were as follows:
• Dr. David prescribed 30 (5 mg) Perco-cet pills; 50 (7.5 mg) Percocet pills; 180 MS Contin (30 mg) pills; 1,624 OxyContin (10 mg) pills; 3,930 Oxy-Contin (20 mg) pills; 10,255 OxyContin (40 mg) pills; 2,543 OxyContin (80 mg) pills; 3,408 Vicodin pills; 31 Tussionex pills; 5,427 Xanax pills; 180 Adipex-P pills; and 3,280 Valium pills. Converting the weights of those pills to marijuana, as provided in U.S.S.G. § 2D1.1, *702resulted in a marijuana equivalent of 4,756.59 kilograms of marijuana.
• Dr. Randy prescribed 40 Percocet (5 mg) pills; 270 MS Contin (30 mg) pills; 231 OxyContin (10 mg) pills; 1,697 OxyContin (20 mg) pills; 3,488 Oxy-Contin (40 mg) pills; 6,010 OxyContin (80 mg) pills; 5,033 Vicodin pills; 24 Tussionex pills; 4,067 Xanax pills; 1,305 Adipex-P pills; 5,285 Valium pills; and 148 Triazolam pills. The marijuana equivalent of these pills was 4,409.6 kilograms of marijuana.
After adding a two-point enhancement to each doctor’s offense level for the use of a special skill, see § 2D1.1 app. note 8, the probation officer assigned an offense level of 36 for each defendant. Paired with each one’s criminal history category of I, both calculations resulted in an advisory Guidelines range of 188 to 235 months’ incarceration. The district court adopted those calculations in full and explained that it found both doctors responsible for all of the prescriptions in all of the files “because the Court finds by clear and convincing evidence that a conspiracy existed in that the prescriptions were written in furtherance of that conspiracy, and that both of these doctors were members of the conspiracy.” The court found that the Doctors jointly owned and operated their practice, shared profits, and participated together in the treatment of more than half of the 98 patients whose files were examined. Each defendant therefore was held responsible not only for the prescriptions that he wrote, but also for those that his co-defendant wrote.
Relying on those findings, the district court sentenced Dr. Randy to the statutory maximum, 60 months’ imprisonment, for his single count of conviction (count 10). Dr. David received a sentence of 188 months’ imprisonment on count 5, a term of 60 months on each of counts 10-12, and 120 months on the health fraud counts, 20-21, all to be served concurrently. Each doctor was also ordered to pay a special assessment and a fine: $100 and $40,000 for Dr. Randy; $600 and $60,000 for Dr. David. Without the relevant conduct enhancements, Dr. Randy’s base offense level would have been 6; the two-point enhancement for special skill would have increased it to 8, resulting in an advisory sentencing range of 0 to 6 months in prison. Dr. David’s offense level before relevant conduct was 14, which would become 16 with the special-skill enhancement, for an advisory sentencing range of 21 to 27 months in prison.
B
Section 1B1.3 defines as relevant conduct “all reasonably foreseeable acts and omissions of others in furtherance of the jointly undertaken criminal activity.” This includes conduct of which the defendant was acquitted, see United States v. Watts, 519 U.S. 148, 156, 117 S.Ct. 633, 136 L.Ed.2d 554 (1997), and conduct for which he was never charged, see United States v. Anderson, 517 F.3d 953, 963 (7th Cir.2008). Nevertheless, the relevant conduct must be unlawful. United States v. Frith, 461 F.3d 914, 917-18 (7th Cir.2006); United States v. Schaefer, 291 F.3d 932, 937-40 (7th Cir.2002) (holding that it is “not enough” to show that defendant’s conduct is “disreputable and unethical”; to qualify as relevant conduct under the Guidelines, the conduct must be unlawful). In calculating relevant conduct, the sentencing court must make its findings by a preponderance of the evidence. The sentencing record before us, however, does not distinguish between conduct that is unlawful and conduct that is the result of mistake or inadvertence. Inattention or negligence could lead a doctor to prescribe medication that is “medically unnecessary”; there is *703nothing necessarily criminal about such behavior. The PSRs did not offer any explanation why the prescriptions in the 98 files were not merely unnecessary, but indicative of illegal drug-pushing. They simply take the data in the Government’s spreadsheets and duplicate the figures as the findings of the probation officer.
At several points in the proceedings, the district court made remarks suggesting it was confused or uncertain about the role that civil standard-of-care evidence should play in the relevant conduct determinations. For example, at the pretrial stage, the Government moved in limine to prohibit the defense from introducing testimony that other physicians had treated the Doctors’ patients in the same way as the defendants had treated them. The Government argued that the other doctors “are not capable of making that assessment whether or not these physicians were prescribing outside the scope of medical practice and procedure” because they, unlike Drs. Parran and Barkin, had not evaluated the patient charts. In response, the district court asked, “doesn’t any doctor have the right to say whether or not a patient that they’re treating within their field, whether the doctor beforehand was treating the patient within the standard of care?” (Emphasis added.) The Doctors argue that while the district court may have been correct that any doctor is capable of opining about whether another physician met the standard of care when treating a particular patient, “the problem here is that the government was discussing the criminal standard (‘outside the scope of medical practice’) and the district court responded with the civil standard (‘the standards of care’).”
During the sentencing hearing on September 6, 2006, the district court indicated that it may have relied on deviations from the civil standard of care in determining relevant conduct. The court stated, for instance: “My recollection, and again, I want to take a look at the evidence, was that some of these patients came in and gave an excuse as to why they needed the drug, not that there was actual necessity.” As we have already said, an absence of medical necessity falls short of the criminal standard for prescribing outside the scope of medical practice altogether. We recognize the diligence of the district court in devoting substantial time to review all of the files and testimony from this long, complex trial. Even so, if the court evaluated the evidence with an eye for detecting failures to live up to the civil standard of care, then it clearly erred. On this record, we are not convinced that the district court properly distinguished between the prosecution’s oft-repeated statements about “medically unnecessary” or “careless” prescribing and the applicable criminal standard of prescribing without a legitimate medical purpose. The blurring of that line becomes even more apparent when we examine the transcript from September 28, 2006, which was when the court articulated its findings and announced the defendants’ sentences.
To support its relevant conduct findings, the district court relied almost exclusively on the testimony and conclusions of Dr. Parran. In effect, the court adopted the position of the Government that classified every prescription for the identified drugs in all 98 files as unlawful. The court discussed only 10 of those 98 files at the hearing, noting how, according to Dr. Par-ran’s testimony, those files offered evidence that the drugs dispensed to those patients were not prescribed for a legitimate medical purpose. This nearly-exclusive reliance on Dr. Parran’s testimony is troublesome, especially given his role as a “standard-of-care” expert.
*704Summing up its findings, the district court stated:
The Court has also taken the opportunity to review each of the 98 patient files in evidence. Numerous files contained evidence that the defendants were put on notice by pharmacies, organizations that monitor what prescriptions people get, and other entities regarding the patient’s drug seeking behavior. Troubling to the Court is that the records illustrate that the defendants turned a blind eye to these notices and continued to prescribe controlled substances to these patients without question. Moreover, there is evidence that nearly one-half of the 98 patients whose files were reviewed, came from the State of Michigan. There was evidence that sometimes two or more Michigan patients would travel together to the defendant’s [sic ] medical practice. Of these Michigan patients, most of them were prescribed controlled substances. This is more than coincidental and yet ignored and never acknowledged or considered by the defendants.
This explanation is problematic for two reasons. First, stating that “[n]umerous files” contained evidence suggesting illicit prescribing is not sufficient to sweep every pill in all 98 files into the relevant conduct calculation. The same goes for the court’s account of Dr. Parran’s testimony, during which it made statements such as: “Dr. Parran found many files had red flags that were totally ignored by the defendants”; and “According to Dr. Parran, diagnostic work-ups were present in very few charts and he noticed that even when consultations were ordered, they were rarely ever followed up” (emphasis added). Such statements are too imprecise and indefinite to establish the illegality of all the prescriptions in all of the files.
Similarly, the fact that “most” of the patients from Michigan were prescribed controlled drugs also provides no concrete foundation for including every pill dispensed to a Michigan patient as relevant conduct. The trial testimony revealed that many patients — from Michigan and Indiana alike — complained of and in fact experienced true medical problems. The court’s assumption of a lack of legitimate medical purpose for every prescription in 98 flies after discussing only 10 flies with any specificity was not enough to support its findings.
As for the scarcity in the files of results from diagnostic tests, such as MRIs and CAT scans, we need only repeat the district court’s own observation that, for many patients, such tests were ordered (sometimes repeatedly) and were not completed; for others, records of completed tests were in the patient’s file. Many patients testified that they failed to comply with testing requests from the Doctors because they could not afford the expensive tests like MRIs. See BlueCross BlueShield Ass’n, “How Much Things Actually Cost,” available at http://www.bcbs. com/coverage/basics/cost (stating the “average national cost of an MRI is nearly $2,000,” and for those insured by BlueC-ross BlueShield, “your cost for an MRI” is $378); Three Rivers Endoscopy Center, “Magnetic Resonance Imaging (MRI),” http://www.gihealth.com/html/education/ mrLhtml (“The cost of an MRI study can range from $400 to more than $2,000, with a typical cost being about $800.”).
We also note that while the 98 patient files at issue were part of the trial record, we initially did not receive them as part of the appellate record, because, it seems, the defendants did not have access to them. We obtained them (with some effort) from the Government, which sent us six boxes. Two boxes contained the 98 files; one con*705tained miscellaneous trial exhibits; three contained nothing but rubber-banded stacks of prescription forms. A prescription form contains only scanty information, usually no more than drug name, dose, date, doctor signature. Very little (if anything) about a prescription form indicates whether it was written “without any legitimate medical purpose.” To the extent that the Government’s spreadsheets and relevant conduct calculations relied on the prescription forms, a better explanation of why any reliable information could be gleaned from them and how it fit into the ultimate decision about relevant conduct was necessary.
The final troublesome part of the sentencing record is the district court’s failure to address evidence tending to suggest a legitimate medical purpose for several prescriptions in the 98 files. For example, the district court specifically inquired during the first date of the sentencing hearing “whether there was any evidence in the trial of the Chubes’ lowering their patient’s [sic ] dosages of OxyContin.” In response to that inquiry, the defendants produced excerpts from 19 patient charts showing reductions in each patient’s dosage. Many of those decreases were accompanied by chart notes stating that the doctor was weaning the patient from OxyContin in an effort to avoid tolerance or addiction. These notes are reinforced by the testimony of several patients, who stated that the Doctors seemed concerned for the patients’ well-being and at times were working to wean the patient off of a strong drug being prescribed to avoid dependence. Any legitimate prescriptions must be deducted from the pill totals before a final determination of relevant conduct is possible.
IV
Thus, while we uphold the Doctors’ convictions, we must remand for resentencing. To establish relevant conduct, the Government bears the burden of showing that a particular prescription was dispensed with no legitimate medical purpose. Presenting only a spreadsheet or a prescription form filled out by one of the defendants is insufficient. This is not a situation, moreover, in which the district court may rely on sampling or extrapolation. Here, unlike other cases we have reviewed under § 841(a), the drug quantity used at sentencing was expressly based on the 98 patient files and the finite set of prescriptions contained within them. These are not defendants who, from a period of “y” to “z,” were dealing drugs on the street to an unclear number of people on an unknown number of occasions. Compare United States v. Noble, 299 F.3d 907, 911 (7th Cir.2002) (“A judge has leeway to extrapolate quantities from witnesses’ [sic] statements of minimum sales over several occasions.... ”); United States v. Durham, 211 F.3d 437, 444-45 (7th Cir.2000) (“[I]t is also permissible for a court to take witness’ [sic ] estimates of the amount of drugs they purchased and multiply that by the minimum quantity sold on each occasion, as well as extrapolate drug quantities from the amount of money used to purchase the drugs.”); United States v. Gaines, 7 F.3d 101, 103-06 (7th Cir.1993); United States v. Martz, 964 F.2d 787, 790 (8th Cir.1992).
In this case, a defined set of concrete data formed the sole basis for determining the quantity of illegally prescribed drugs. For a prescription to be included in relevant conduct, the court must evaluate the facts surrounding that particular prescription and explain why those facts render it unlawful. Generalizing from “numerous” files will not suffice. When the district court revisits relevant conduct on remand, it must explain its findings with respect to *706each patient and make a reasoned determination whether or not the Government has carried its burden of establishing that each prescription was dispensed outside the scope of medical practice and without a legitimate medical purpose.
The convictions of David Demaret Chube II and Charles Randall Chube are AffiRmbd. Both sentences are Vacated, and we ReMand for resentencing each defendant in accordance with this opinion.