EVIDENCE LIBRARY
Does Drinking Water Between Alcoholic Drinks Prevent a Hangover? A 13-Man Trial
A 2026 crossover trial tested extra water with a fixed ethanol dose. Learn what breath ethanol, acetaldehyde, hangover ratings and DSST actually showed.
Drinking extra water between alcoholic drinks did not measurably reduce next-morning hangover symptoms or change exhaled ethanol and acetaldehyde profiles in a small controlled trial when the alcohol dose was kept the same. In a randomized two-period crossover experiment, 13 healthy Japanese men consumed the same substantial dose of sake on two occasions, once with and once without an additional prescribed amount of water during the drinking hour. The researchers detected no statistically significant differences between the conditions in their measured breath-alcohol and breath-acetaldehyde trajectories or six self-rated symptoms. A brief digit-symbol task produced a worse result with added water immediately after drinking and inconclusive next-morning differences. The study does not establish that water has no possible benefit for everyone, nor does it measure forensic whole-blood alcohol concentration, actual driving ability, or accident causation.
This distinction matters whenever a driver, witness, defendant, or expert argues that alternating alcohol and water necessarily makes a drinking episode safer. Hydration, breath chemistry, perceived symptoms, cognitive test results, and impairment at a particular moment are different evidentiary questions. The original experiment by Imai and colleagues, published June 15, 2026, addresses only certain parts of that chain: Imai et al., Water intake during drinking does not alleviate hangover symptoms, Frontiers in Pharmacology, DOI: 10.3389/fphar.2026.1852528.
What did the water-versus-alcohol experiment actually test?
The question was not whether someone who replaces an alcoholic beverage with water consumes less alcohol. Investigators held the ethanol dose constant and added water in one condition. This separates two real-world practices that are frequently confused: drinking the same amount of ethanol plus water, and reducing ethanol consumption by alternating with nonalcoholic drinks. The study examined the first, not the second.
Each volunteer drank Japanese sake containing 15% alcohol by volume, dosed to provide 1.3 g of pure ethanol per kilogram of body mass over one hour. During the comparison visit, the participant drank an additional 15 mL of mineral water per kilogram at intervals during the same hour. A participant weighing 70 kg would therefore have a prescribed additional water amount of 1,050 mL; this example is a dose calculation, not a reported individual measurement. Drinking took place from approximately 9 p.m. to 10 p.m. with a standardized dinner beforehand. Visits were separated by at least seven days.
The design randomized which condition came first. It was a crossover trial, so the same participant provided measurements under both protocols. It was not blinded: a person can plainly notice drinking an additional liter or more of water. All participants also had restricted access to bottled water after the drinking period and the following morning under the protocol. Accordingly, the comparison was about extra water during alcohol administration, not complete dehydration versus hydration.
Who participated, and how broadly do the findings apply?
Sixteen men consented; two withdrew and one was excluded after being unable to consume the prescribed water quantity. The analysis therefore included 13 completers. Their mean age was 36.2 years (standard deviation 9.6), mean body weight was 71.5 kg (standard deviation 8.2), and mean body mass index was 24.0 (standard deviation 2.0). Reported sake consumption averaged 749 mL per session, with a range of 598–967 mL. The added-water quantity averaged 1,072 mL, ranging from 855 to 1,383 mL.
Selection was restrictive. The volunteers were healthy Japanese adult men with the wild-type ALDH2 *1/*1 genotype; the researchers excluded certain medical conditions and medication use. They did not test women, persons with reduced-function ALDH2 genotypes, older adults with significant illness, or people whose eating and water intake followed ordinary uncontrolled social patterns. The investigators characterized this as an exploratory study, without a definitive sample-size calculation for small effects. Thus, it would be incorrect to generalize a null group comparison to every sex, genotype, dose, beverage, or population.
What biological specimens and instruments supplied the alcohol findings?
Exhaled breath, not collected forensic whole blood, was the chemical measurement matrix. Participants rinsed their mouths with water, provided end-expiratory samples into collection bags, and investigators analyzed 5 cubic centimeters of sampled gas by sensor gas chromatography. The reported detection limits were 0.2 parts per million (ppm) for breath ethanol and 5 parts per billion (ppb) for breath acetaldehyde. These are gas-phase concentrations with different analytes and units; they are not two interchangeable blood tests.
Ethanol is the parent alcohol present in the beverage. Acetaldehyde is a metabolite and can also have sources or sampling considerations that must be evaluated. Their breath concentrations are observations of exhaled chemical species under the specified sampling protocol. This trial did not provide a paired forensic venous whole-blood ethanol series measured by headspace gas chromatography. It did not establish an individual blood-to-breath partition ratio, nor validate converting these ppm or ppb values into a particular driver’s statutory blood alcohol concentration. For more background on the matrix issue, see breath alcohol measurement and pharmacokinetics in toxicology.
What did the researchers observe across the 15-hour sampling window?
Researchers sampled before drinking, immediately and one hour after the drinking period, and at scheduled assessments following approximately six hours of sleep, extending to 15 hours after drinking ended. Their statistical approach used linear mixed-effects models accounting for the beverage condition, study period, and repeated participation. Significance testing was exploratory, with a nominal two-sided 5% threshold and no adjustment for multiple comparisons.
They reported no statistically significant water-versus-no-added-water difference in the breath ethanol or acetaldehyde trajectories at the assessed time points. The numerical mean breath acetaldehyde values just after drinking and one hour later appeared somewhat higher with water, but that early contrast was not statistically significant. By the morning assessments, breath profiles were described as very similar. The paper does not give a sound basis for treating an isolated plotted mean or visually estimated graph interval as a validated person-specific alcohol elimination rate.
A non-significant difference does not demonstrate exact equivalence. With 13 analyzed volunteers, the trial had limited power to detect modest effects. It is accurate to say that this study did not detect a difference under its conditions; it is not accurate to say extra water has been proved incapable of modifying any alcohol kinetic parameter in any circumstance.
What happened on the digit-symbol psychomotor task?
The investigators also administered the Digit Symbol Substitution Test (DSST), recording the number of correct symbol-to-number substitutions achieved within one minute. This is a standardized laboratory performance task involving processing speed, attention, and visual-motor responding. It is not an on-road driving assessment, a validated legal-intoxication cutoff, or a complete battery of neuropsychological functions.
Immediately at the end of drinking, the change in DSST score from baseline was lower in the additional-water condition than the no-extra-water condition (p = 0.01). The authors discussed abdominal fullness and discomfort as one possible contributor, not as a proved mechanism. At seven and eight hours after drinking, the water-condition point estimates favored higher DSST scores, but the published analysis did not characterize the next-morning differences as a reliable improvement: it reports p = 0.05 at seven hours and p = 0.08 at eight hours and cautions against a positive interpretation.
The reported seven-hour value lies at the nominal significance boundary and the study did not adjust across its multiple time-point comparisons. A responsible summary therefore does not turn that value into proof of next-morning cognitive protection, particularly when the primary authors describe the morning effects as not statistically significant. Nor should the early p = 0.01 result be converted into a conclusion that water causes functional driving impairment. It is a task difference in a small open-label experiment, potentially affected by discomfort and testing conditions.
Did additional water relieve headache, nausea, sleepiness, or subjective intoxication?
Six subjective domains were assessed on 100-mm visual analogue scales: facial flushing, headache, nausea, concentration, sleepiness, and mood elevation. These are self-reported feelings, not laboratory ethanol concentrations or independently measured task performance. Consistent with an intoxication period followed by a symptomatic morning, the investigators described the expected pattern of symptoms across the sessions.
For the comparison that matters here, none of the six symptom outcomes showed a statistically significant difference between drinking the fixed sake dose with additional water and drinking it without that additional in-hour water. The conclusion is about these symptom ratings in these 13 analyzed participants, at the times and dose studied. It does not mean that every participant felt exactly the same, that ordinary hydration has no medical relevance, or that a person who reports fewer symptoms must be fit to drive.
Evidence table: what was measured, and what can it prove?
| Study element | Primary study result or method | Forensic boundary |
|---|---|---|
| Design and dose | 13 completers; randomized sequence crossover; 1.3 g/kg ethanol in 15% sake over one hour, with versus without 15 mL/kg extra water | Same ethanol exposure was intended in both sessions; not a test of drinking less alcohol |
| Breath ethanol | End-expiratory sensor-GC gas-phase ethanol; detection limit 0.2 ppm; no statistically significant condition difference reported | Not a directly collected or measured forensic whole-blood BAC |
| Breath acetaldehyde | Gas-phase acetaldehyde; detection limit 5 ppb; early mean contrasts not significant | Indirect metabolite finding, not a validated individualized impairment indicator |
| Immediate DSST change | Water condition lower than no-extra-water at 0 hours, p = 0.01 | One processing-speed task, not road driving or a causal diagnosis |
| Next-morning DSST | Seven-hour p = 0.05; eight-hour p = 0.08; authors did not establish a significant morning benefit | Exploratory, unadjusted time-point tests; cannot classify one person’s fitness |
| Subjective VAS symptoms | Six domains; no statistically significant condition effects | Self-report is neither chemical verification nor objective impairment classification |
Why the distinction between extra water and less alcohol matters
The intervention did not replace sake with water. Its fixed ethanol challenge was designed to isolate a water addition from a reduced alcohol dose. In ordinary life, a person who alternates an alcoholic drink with water may also drink fewer alcoholic drinks, drink more slowly, eat differently, or stop sooner. These behavioral changes could affect exposure independently of any direct water effect. The experiment did not randomize people to those real-world patterns, and its negative comparison cannot adjudicate them.
Drinking water also must not be confused with accelerating ethanol metabolism. Changes in total fluid volume are not the same as a demonstrated rise in hepatic clearance. The study did not directly quantify individualized liver clearance or test a validated method of speeding elimination. A claimed correction of a measured breath or blood alcohol result on the ground that someone consumed water would require separate, appropriate kinetic evidence rather than a generic assumption about hydration.
The difference between a drink’s concentration and the total dose also matters. Adding water could dilute beverage contents in the stomach while leaving ethanol grams unchanged. That does not automatically imply a proportional change in peak concentration, elimination slope, or the concentration later measured in a forensic specimen. Timing, gastric handling, and interindividual variation can all matter; this study was not designed to provide a conversion formula for a case reconstruction.
Which limitations prevent strong individual or causal claims?
The small n = 13 per-protocol sample, all-male ALDH2 genotype restriction, open-label condition, unusually prescribed water quantity, high fixed sake dose, and two-visit laboratory environment limit external validity. The study did not separately measure dehydration with a validated clinical endpoint, sleep quality, roadway behavior, collisions, injury risk, or long-term outcomes. The added-water volume was large enough that some participants reported bloating, with one unable to complete that condition. Such effects could affect self-report or brief psychomotor testing.
There were multiple breath, symptom, and psychomotor comparisons without multiplicity correction, and there was no definitive equivalence or noninferiority design. A lack of statistical significance is therefore not affirmative proof that the true average effect is exactly zero. The authors’ account of possible early absorption changes was mechanistic interpretation, not a directly proven individual transport mechanism. The selected sake beverage also contains nonethanol constituents; one cannot assume identical results for beer, wine, spirits, or cocktails.
The paper disclosed corporate funding from the sake producer, and several authors were employed by the producer. The publication states the funder did not direct study design, collection, analysis, interpretation, writing, or submission. Those disclosures should be read alongside the research methods and results rather than used as a substitute for scientific appraisal.
Evidence table: which litigation propositions remain untested?
| Proposed case-specific inference | Was it established in this experiment? | Why the evidence is insufficient |
|---|---|---|
| Adding water necessarily lowers forensic whole-blood BAC | No | No matched venous whole-blood ethanol series or validated individual water correction |
| Water eliminates next-day symptoms for a given person | No | Selected group, six VAS ratings, no detected average effect, incomplete individual prediction |
| Next-morning task performance is reliably protected | No | Exploratory DSST comparisons at seven and eight hours, no multiplicity adjustment or driving endpoint |
| Drinking water proves a driver was safe or caused no crash | No | No individual driving observations, collision analysis, or case-specific causation evidence |
How should the study be used in DUI, workplace, and injury litigation?
For a DUI or occupational-safety dispute, the precise supported statement is narrow: at a matched experimentally administered ethanol dose, adding the specified water during one hour of drinking did not yield statistically distinguishable breath ethanol/acetaldehyde profiles or measured next-morning self-rated symptoms in this small selected sample. The study does not establish an individual’s whole-blood BAC at the time of driving, nor can it validate a calculation from water intake to legally relevant BAC.
For performance testimony, the DSST is direct evidence of a particular observed laboratory task, not an individualized prediction of lane control, divided attention in traffic, reaction to a hazard, or negligence. A single statistical result is not a universal impairment threshold. A person may report feeling normal but have an unmeasured performance deficit, or report discomfort that does not establish an alcohol-related driving deficit. Neither subjective state alone nor a breath metabolite curve proves accident causation.
If counsel asks whether a driver was “protected” by water, the defensible response is to identify the actual alleged protection—lower dose, lower measured BAC, fewer symptoms, better task performance, or avoidance of a collision—and then ask whether that endpoint was measured and validated for that person. Specimen collection timing, analytic method, drinking chronology, clinical observations, and task-specific evidence remain separate. See the forensic toxicology evidence library for additional context on evaluating measurement-based claims.
Frequently asked questions about water, hangovers, and alcohol testing
Does a glass of water reduce a breathalyzer reading?
This trial did not detect a significant change in its laboratory breath ethanol time course when extra water was added to the same ethanol dose. It did not examine every breathalyzer, every drinking pattern, or mouth-alcohol artifact, and it does not justify adjusting an individual’s evidentiary breath result.
Does the study prove water cannot prevent a hangover?
No. It found no statistically significant improvement in the six tested subjective symptoms among 13 selected male completers under a high fixed sake dose. Smaller benefits, other populations, or effects mediated by consuming less alcohol were not ruled out.
Were blood alcohol concentrations directly measured?
No paired forensic venous whole-blood ethanol concentration series was reported for this comparison. Ethanol and acetaldehyde were analyzed from exhaled gas. Breath ppm and ppb are not whole-blood BAC units or a validated individual conversion in this trial.
Did water make psychomotor performance worse?
The water condition had a lower baseline-adjusted one-minute DSST score immediately after drinking (p = 0.01). That one observation does not establish a general impairment effect or identify the cause of the difference; the study discussed possible abdominal discomfort.
Was next-morning reaction time improved?
The study measured correct responses on the DSST rather than a validated on-road reaction-time endpoint. Morning comparisons at seven and eight hours were not treated by the investigators as convincing evidence of benefit; the reported values were p = 0.05 and 0.08, without multiplicity correction.
Can a person feel sober and still have relevant impairment?
Yes, subjective ratings and actual task performance are different constructs. This experiment did not validate a rule linking either person’s self-rating or water intake to their driving competence at a specific event.
Can experts use this paper to calculate a driver’s BAC or prove an accident was caused by alcohol?
No. It provides no validated individual water-to-BAC correction, no matched forensic whole-blood assay series, no crash investigation, and no causal model for a particular injury. Case-specific conclusions require independently relevant evidence and appropriately limited testimony.
Bottom line and primary research citation
With ethanol grams held constant, extra water taken during the drinking hour did not produce a demonstrated reduction in measured breath ethanol, breath acetaldehyde, or six subjective hangover symptoms in a 13-man randomized crossover trial. The results are a limitation on broad claims that a “water chaser” reliably neutralizes drinking, not proof that hydration never matters. Immediate DSST worsening and inconclusive next-morning task contrasts require their own careful interpretation. The correct scientific separation remains: an exhaled-analyte measurement is not an individual forensic whole-blood BAC; a symptom score is not objective driving performance; and neither is proof of legal impairment or incident causation.
Original primary source: Imai H, Kushio S, Matsuura K, Nakamura A, Mori K, Kadowaki M, Oikawa I, Uemura N. Water intake during drinking does not alleviate hangover symptoms. Frontiers in Pharmacology. Published June 15, 2026;17:1852528. doi:10.3389/fphar.2026.1852528.