Snap Shot
- The question: How accurately do people report how much they drank and how intoxicated they were, and what follows for any account of drinking, and any blood alcohol estimate, built on what a complainant, an accused, or a witness says?
- What the research shows: In a field study of 81 young adults in one city that compared reports with observers’ counts, reports were accurate on average for nights of up to about eight drinks, and underestimation grew as drinking increased. In one U.S. national survey, self-reports accounted for about 31% of per capita alcohol sales before adjustment, though some of that gap reflects alcohol that is sold but not drunk; it is a population finding, not a statement about any individual. People commonly misjudge what counts as a drink, and bar and home pours often exceed a standard drink. In one bar study of 454 patrons already at or above 0.08%, about 60% judged themselves below that level, and in a scenario study of college students about 20% underestimated intoxication; drink-based BAC estimates explain only about half the variation in measured breath alcohol.
- What it does not show: The research does not show that any particular person’s account is too low or too high, or that a person who reports few drinks is being untruthful. Some findings point toward underreporting and others toward overestimation, and the direction of error in a single account depends on the facts.
- The law: Testimony that relies on, or explains the limits of, reported drinking is governed by Rule 702 (as amended in 2023), Daubert, Kumho, Mil. R. Evid. 702 and Houser, and Ariz. R. Evid. 702; an Arizona appellate court treated gaps in a driver’s eating and drinking history as a matter of weight for a retrograde extrapolation (State ex rel. Montgomery v. Miller, App. 2014). Experts may explain the general research but may not opine on whether a particular witness is truthful (State v. Lindsey, 1986).
- Issues for each side: Either side may legitimately point to the general unreliability of drink counts; both should apply that point to every party’s account, look for receipts, video, and other records that can check it, and avoid treating the research as evidence that a particular person lied.
Most cases involving alcohol begin with a question about how much someone drank. Investigators ask complainants, accused persons, and witnesses; experts convert the answers into blood alcohol estimates; and the answers are compared with one another and with test results. Yet an account of drinking is a memory, formed while drinking and recalled later, often after a night the person does not fully remember. It depends on what the person noticed, how the person counted, what the person thought a “drink” was, and how willing the person was to disclose it.
This paper reviews what is known about the accuracy of self-reported drinking: field studies comparing reports with observation, population surveys compared with sales, research on drink size and pouring, studies of how well people estimate their own intoxication, and the accuracy of BAC estimates built from reported drinks. It then addresses the limits of that research and the law. An older page on this topic listed abstracts; this paper replaces it. The calculations that turn drink counts into BAC estimates are addressed in Retrograde Extrapolation and the Limits of the Widmark Formula.
How Accurately Do People Report a Night’s Drinking?
The most direct test of accuracy compares what people say they drank with what observers saw. In a study in Perth, Australia, peer researchers accompanied 81 young adults aged 18 to 25 on nights out and recorded their drinks; one to two days later, participants reported how many drinks they had consumed. For sessions of about eight drinks or fewer, reports were accurate on average, but as drinking increased beyond that level, participants underestimated by increasing amounts (Northcote & Livingston, 2011). The study is small and covers one city’s nightlife, but it is one of the few to verify reports against observation in a natural setting, and its pattern, accuracy at moderate levels and growing underestimation at heavy levels, is the one most relevant to cases involving heavy drinking.
Calendar-based interviews, in which a person reconstructs each day’s drinking with the help of personal landmarks, have shown acceptable reliability for recent drinking among non-dependent drinkers (Sobell et al., 1988). Reliability, though, means consistency between repeated reports; it does not establish that either report is accurate, and the method assumes the person remembers the drinking. For hours a person does not remember, the person can report only what others have said or what records show, a problem addressed in Reconstructing a Blackout Period from Digital and Documentary Evidence.
What Surveys Miss
Population surveys show that self-reports, taken together, account for much less alcohol than is sold. In a U.S. national survey of more than two million adults, self-reported consumption accounted for about 31% of per capita alcohol sales before adjustment (Esser et al., 2022). Researchers using a “yesterday” method, which asks in detail about the previous day’s drinking, have estimated who underreports. In a Canadian national survey of 43,371 people, underreporting was greater among younger drinkers, was greater among lower-risk than higher-risk drinkers, and did not differ between men and women (Stockwell et al., 2014). A comparison across Australia, Canada, England, and the United States found that people who described themselves as infrequent drinkers substantially underreported how often they drank (Stockwell et al., 2016).
These figures describe populations answering survey questions about their usual drinking, not a person reporting a specific night to an investigator. Sales include alcohol that is spilled, stockpiled, or consumed by visitors, and survey samples underrepresent the heaviest drinkers. They nonetheless establish that underreporting is common, that it is not confined to one group, and that, at the population level, the error runs predominantly toward reporting too little.
What Counts as a “Drink”?
Counting drinks assumes a shared unit. In the United States a standard drink contains about 14 grams of pure alcohol, roughly the amount in 12 ounces of regular beer, 5 ounces of table wine, or 1.5 ounces of 80-proof spirits (National Institute on Alcohol Abuse and Alcoholism [NIAAA], n.d.), but people rarely drink in those units. In a study of 133 college students asked to define and free-pour standard drinks, students overestimated the volume of a standard serving of wine and spirits and poured more than a standard drink; when they were shown the correct volumes, their reports of recent drinking rose, suggesting their original reports had been too low (White et al., 2005). Commercially poured drinks also vary. When researchers purchased and measured drinks at 80 bars and restaurants in Northern California, the average glass of wine contained 43% more alcohol than a standard drink, the average draught beer 22% more, and the average mixed drink 42% more, while shots were about equal to a standard drink (Kerr et al., 2008). More recent work has used students’ free-pouring errors to adjust self-reported consumption when studying its consequences (Hussong & Traver, 2026).
The practical result is that “four drinks” can mean four standard drinks or the alcohol equivalent of six or more. A report of the number of glasses or cups, without their size and strength, leaves an estimate uncertain, and the research suggests the error more often runs toward too little alcohol than too much. The same uncertainty applies to a witness’s account of what another person drank, which is also a memory formed while the witness was often drinking.
How Well Do People Judge Their Own Intoxication?
People’s sense of how intoxicated they are is also imprecise. In a field study of 454 bar patrons, all of whom had breath alcohol at or above 0.08%, 60.4% judged their level to be below 0.08%; women and patrons under 21 were more likely than others to estimate below that level (Martin et al., 2016). Because every participant was at or above 0.08%, the design could detect underestimation but not overestimation. In a study of 3,707 college students asked to estimate intoxication from drink scenarios, about 20% underestimated, women made more underestimation errors in scenarios with four to five drinks and men in scenarios with six to seven, and women who typically drank heavily at parties were more likely to err at moderately high drink counts (Mallett et al., 2009). A review of training studies concluded that moderate drinkers can learn to estimate their BAC with feedback but that people with alcohol dependence, likely because of tolerance, do poorly (Aston & Liguori, 2013).
Feeling less intoxicated than one is also has a pharmacological basis. People generally feel less intoxicated on the falling part of the BAC curve than at the same level while it is rising, although objective skills are generally as impaired or more so (Holland & Ferner, 2017; see Acute Tolerance and the Mellanby Effect). A person’s report that he or she “did not feel drunk,” whether the person is a complainant, an accused, or a witness, is therefore limited evidence of how impaired the person was, because people judge their own intoxication imprecisely in both directions; it should be considered together with observed behavior, drinking, and testing.
From Drink Counts to BAC Estimates
Drink counts become BAC estimates through equations that add body weight, sex, and time. In a field validation, estimated BACs calculated from the drinks reported by 109 people recruited near bars agreed closely across several equations but explained only about 55% of the variance in measured breath alcohol, overestimated on average, and fit less well as intoxication rose (Hustad & Carey, 2005). In a laboratory drinking-game study with 40 participants, formula-based estimates tended to run higher than breath readings, more so for women (Silvestri et al., 2013).
These findings show that the error in a drink-based estimate is not one-directional. Underreported drinks and larger-than-standard pours push an estimate down; the equations themselves, at least in these studies, tended to push it up. Which effect dominates in a particular case depends on the facts. On average, women reach higher BACs than men from the same number of drinks, mainly because of differences in body water (Mumenthaler et al., 1999); body size, food, drinking speed, tolerance, and timing also matter, so equal drink counts do not imply equal intoxication for two people of different sex or build, in either direction.
Hypothetical Example. Consider a hypothetical case in which two people each tell investigators they had “about five drinks” at a party. A host’s photographs show both drinking from large plastic cups filled from a punch bowl of unknown strength, and a receipt shows that the host bought several bottles of spirits. Neither person’s account can be taken as a count of standard drinks, and neither can be assumed to be more accurate than the other because of the person’s role in the case. The research suggests the reports are more likely to understate than overstate the alcohol consumed, but by how much is unknown; an estimate built on them should carry a wide range, and the photographs, receipt, and other witnesses’ observations may narrow it.
Limits of the Evidence
The few studies that verify reports against observation are small, involve young adults in particular nightlife settings, and examine moderate to heavy but not extreme drinking. Survey comparisons with sales describe populations, include alcohol not consumed by respondents, and say little about a single night. Studies of drink definitions and pours come mainly from U.S. college students and Northern California establishments. Intoxication-estimate studies used bar patrons already over the legal limit or hypothetical scenarios. Sex and gender were examined in several studies: underreporting did not differ by gender in the Canadian survey (Stockwell et al., 2014), women were more likely to underestimate their BrAC among bar patrons (Martin et al., 2016), the pattern of underestimation errors differed by gender in scenario judgments (Mallett et al., 2009), and formula-to-breath discrepancies were larger for women in one small study (Silvestri et al., 2013); none of these differences shows how accurately a particular woman or man reports drinking. None of the research addresses deliberate concealment separately from honest error, and none can determine whether a particular account is accurate.
Questions to Ask About This Research
When Northcote and Livingston (2011) is cited to suggest that a person drank more than reported, counsel can ask whether the reported amount was in the range where the study found accurate reports (about eight drinks or fewer) or above it, and whether 81 young adults in one city’s nightlife resemble the person. When survey underreporting is cited (Esser et al., 2022; Stockwell et al., 2014), counsel can ask whether a population comparison with sales bears on one person’s account of one night. When Martin et al. (2016) is cited, counsel can ask whether a sample limited to people already over 0.08% could have detected overestimation. When drink-size research is cited (White et al., 2005; Kerr et al., 2008), counsel can ask what is actually known about the containers, pours, and strength in the case. And when a BAC estimate is built on a reported drink count, counsel can ask how the estimate would change if the count or drink sizes were off in either direction (Hustad & Carey, 2005). These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
Self-Reported Drinking and the Law
Federal and Military Standards
Expert testimony that relies on reported drinking, or that explains the limits of such reports, is governed by Federal Rule of Evidence 702, as amended December 1, 2023, which requires the proponent to show that it is more likely than not that the testimony rests on sufficient facts or data, reflects reliable principles and methods, and reflects a reliable application of those methods to the case. Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) makes the court a gatekeeper, and Kumho Tire Co. v. Carmichael (1999) extends that obligation to all expert testimony. Whether a drink count supplied by a party is a sufficient basis for a BAC estimate is the kind of “sufficient facts or data” question the 2023 amendment emphasizes. An expert may explain the general research on self-report accuracy but may not tell the factfinder that a particular witness’s account is true or false. Military Rule of Evidence 702 tracks the federal rule, and the 2023 amendment took effect in the military rule in 2025 (Joint Service Committee on Military Justice, 2025); practitioners should confirm current text. Military judges also apply the factors of United States v. Houser (1993), and testimony that amounts to an opinion on a witness’s truthfulness is impermissible vouching (United States v. Brooks, 2007).
Arizona
Arizona Rule of Evidence 702 adopted the federal formulation in 2012 and was amended effective January 1, 2024, to track the 2023 federal amendment (Ariz. Sup. Ct. Order R-23-0004). General educational testimony that satisfies Rules 702 and 403 is permitted (State v. Salazar-Mercado, 2014), but experts may not opine on whether a particular witness is truthful (State v. Lindsey, 1986, 149 Ariz. at 475; State v. Moran, 1986). In State ex rel. Montgomery v. Miller (App. 2014), the Court of Appeals vacated a trial court order that had excluded a retrograde extrapolation because the driver’s eating and drinking history was unknown; it reasoned that the expert used conservative assumptions and reported a range, and that the missing history went mainly to weight, to be tested by cross-examination and contrary evidence. The decision concerned a calculation from a blood test, not a drink count, but it shows how gaps in a drinking history tend to be treated. Arizona’s general credibility instruction directs jurors to consider, among other factors, the quality of the witness’s memory and whether the witness was contradicted by anything the witness said or wrote before trial (RAJI (Criminal) Standard 15), factors that apply to accounts of drinking as to any other testimony.
Special Issues in Legal Practice
The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.
Special Issues for Prosecutors
The research supports several propositions prosecutors can legitimately present. Underestimation of drinking grows at heavy consumption, poured drinks often exceed a standard drink, and intoxicated people often underestimate their own intoxication, so an accused’s or a witness’s low drink count does not establish light drinking. A complainant or an accused who reports drinking less than receipts or video later show may be giving an honest but inaccurate account; underreporting is common across groups and is not, by itself, evidence of fabrication. Receipts, tabs, video, and other witnesses can test any account of drinking, and gathering them early helps every party.
The research also counsels caution. The same findings apply to a complainant’s account, and an argument that an accused’s drinking was underreported implies that a complainant’s may be as well. In two small validation studies (109 and 40 participants), equation-based estimates tended to exceed measured breath alcohol, while underreporting of drinks, which is common at high consumption, pushes estimates the other way; the net direction depends on the case. Survey underreporting describes populations, not one person’s account of one night, and testimony that a particular person “must have” drunk more than reported goes beyond the research.
Special Issues for Defense
The research supports several propositions defense counsel can legitimately present. Self-reported drinking is a memory, and an estimate of intoxication built on it inherits its uncertainty; formula estimates explained only about half the variance in measured breath alcohol in field validation (Hustad & Carey, 2005). Accounts of another person’s drinking are subject to the same limits, especially when the witness was also drinking. It is legitimate to ask who supplied a drink count, when, after what discussion, and how drink sizes were determined, and to offer expert testimony on these general principles.
The research also counsels caution about the limits that apply to the defense. In surveys and field studies, self-reports at heavy consumption tend to understate drinking, which means that any party’s account of moderate drinking may understate the alcohol consumed; in two small validation studies (109 and 40 participants), equation-based estimates tended to exceed measured breath alcohol, so the net direction of error in an estimate depends on the case. A complainant’s or an accused’s inaccurate drink count is consistent with ordinary memory error and does not, by itself, show whether the rest of the account is reliable or unreliable. The research cannot show that a particular person misstated his or her drinking, and an argument that rests on it for one party’s account is equally applicable to the other’s.
This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.
References
Aston, E. R., & Liguori, A. (2013). Self-estimation of blood alcohol concentration: A review. Addictive Behaviors, 38(4), 1944–1951. https://doi.org/10.1016/j.addbeh.2012.12.017
Esser, M. B., Sherk, A., Subbaraman, M. S., Martinez, P., Karriker-Jaffe, K. J., Sacks, J. J., & Naimi, T. S. (2022). Improving estimates of alcohol-attributable deaths in the United States: Impact of adjusting for the underreporting of alcohol consumption. Journal of Studies on Alcohol and Drugs, 83(1), 134–144. https://doi.org/10.15288/jsad.2022.83.134
Holland, M. G., & Ferner, R. E. (2017). A systematic review of the evidence for acute tolerance to alcohol – the “Mellanby effect.” Clinical Toxicology, 55(6), 545–556. https://doi.org/10.1080/15563650.2017.1296576
Hussong, A. M., & Traver, J. M. (2026). Optimizing measures of drinking to predict alcohol-related consequences. Drug and Alcohol Dependence, 278, Article 112973. https://doi.org/10.1016/j.drugalcdep.2025.112973
Hustad, J. T. P., & Carey, K. B. (2005). Using calculations to estimate blood alcohol concentrations for naturally occurring drinking episodes: A validity study. Journal of Studies on Alcohol, 66(1), 130–138. https://doi.org/10.15288/jsa.2005.66.130
Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf
Kerr, W. C., Patterson, D., Koenen, M. A., & Greenfield, T. K. (2008). Alcohol content variation of bar and restaurant drinks in Northern California. Alcoholism: Clinical and Experimental Research, 32(9), 1623–1629. https://doi.org/10.1111/j.1530-0277.2008.00741.x
Mallett, K. A., Turrisi, R., Larimer, M. E., & Mastroleo, N. R. (2009). Have I had one drink too many? Assessing gender differences in misperceptions of intoxication among college students. Journal of Studies on Alcohol and Drugs, 70(6), 964–970. https://doi.org/10.15288/jsad.2009.70.964
Martin, R. J., Chaney, B. H., Cremeens-Matthews, J., & Vail-Smith, K. (2016). Perceptions of breath alcohol concentration (BrAC) levels among a sample of bar patrons with BrAC values of 0.08% or higher. Psychology of Addictive Behaviors, 30(6), 680–685. https://doi.org/10.1037/adb0000203
Mumenthaler, M. S., Taylor, J. L., O’Hara, R., & Yesavage, J. A. (1999). Gender differences in moderate drinking effects. Alcohol Research & Health, 23(1), 55–64.
National Institute on Alcohol Abuse and Alcoholism. (n.d.). What is a standard drink? Retrieved November 6, 2026, from https://www.niaaa.nih.gov/alcohols-effects-health/what-standard-drink
Northcote, J., & Livingston, M. (2011). Accuracy of self-reported drinking: Observational verification of ‘last occasion’ drink estimates of young adults. Alcohol and Alcoholism, 46(6), 709–713. https://doi.org/10.1093/alcalc/agr138
Silvestri, M. M., Cameron, J. M., Borsari, B., & Correia, C. J. (2013). Examining alcohol and alcohol-free versions of a Simulated Drinking Game Procedure. Journal of Studies on Alcohol and Drugs, 74(2), 329–336. https://doi.org/10.15288/jsad.2013.74.329
Sobell, L. C., Sobell, M. B., Leo, G. I., & Cancilla, A. (1988). Reliability of a timeline method: Assessing normal drinkers’ reports of recent drinking and a comparative evaluation across several populations. British Journal of Addiction, 83(4), 393–402. https://doi.org/10.1111/j.1360-0443.1988.tb00485.x
Stockwell, T., Zhao, J., & Macdonald, S. (2014). Who under-reports their alcohol consumption in telephone surveys and by how much? An application of the ‘yesterday method’ in a national Canadian substance use survey. Addiction, 109(10), 1657–1666. https://doi.org/10.1111/add.12609
Stockwell, T., Zhao, J., Greenfield, T., Li, J., Livingston, M., & Meng, Y. (2016). Estimating under- and over-reporting of drinking in national surveys of alcohol consumption: Identification of consistent biases across four English-speaking countries. Addiction, 111(7), 1203–1213. https://doi.org/10.1111/add.13373
White, A. M., Kraus, C. L., Flom, J. D., Kestenbaum, L. A., Mitchell, J. R., Shah, K., & Swartzwelder, H. S. (2005). College students lack knowledge of standard drink volumes: Implications for definitions of risky drinking based on survey data. Alcoholism: Clinical and Experimental Research, 29(4), 631–638. https://doi.org/10.1097/01.ALC.0000158836.77407.E6
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
State ex rel. Montgomery v. Miller, 234 Ariz. 289, 321 P.3d 454 (App. 2014).
State v. Lindsey, 149 Ariz. 472, 720 P.2d 73 (1986).
State v. Moran, 151 Ariz. 378, 728 P.2d 248 (1986).
State v. Salazar-Mercado, 234 Ariz. 590, 325 P.3d 996 (2014).
United States v. Brooks, 64 M.J. 325 (C.A.A.F. 2007).
United States v. Houser, 36 M.J. 392 (C.M.A. 1993).
Fed. R. Evid. 702.
Mil. R. Evid. 702, 1102.
Ariz. R. Evid. 702.
Ariz. Sup. Ct. Order R-23-0004 (amending Ariz. R. Evid. 702, effective Jan. 1, 2024).
Revised Arizona Jury Instructions (Criminal), Standard Criminal 15 (Credibility of Witnesses).
Related topic papers: Retrograde Extrapolation and the Limits of the Widmark Formula · Alcohol Toxicology for Attorneys: Absorption, Distribution, Elimination, and Stages of Intoxication · Reconstructing a Blackout Period from Digital and Documentary Evidence · Acute Tolerance and the Mellanby Effect · Sex Differences in Alcohol Effects and Blackouts
Related topics from Arizona Forensic Services: Memory in the Courtroom: Alcohol and Memory