One Drink Doesn’t Help You Sleep. Here’s What It Does Instead

You wake up and the two halves of the screen disagree. Sleep stages: normal. Time asleep: normal. Recovery: red, resting heart rate sitting two or three beats above where it usually is. You had two glasses of wine with dinner, finished before nine, slept the same seven hours you always sleep. The obvious conclusion is that the device is being dramatic.

That disagreement is the whole story, and it is better documented than almost anything else in this literature.

Here is why a publication about working with your head is writing about a drink — and here is the limit of it. Alcohol is one of the most widely used sleep aids there is: in a randomly dialled sample of more than two thousand adults, thirteen per cent reported using it to get to sleep, a figure I have from Colrain and Baker’s 2024 commentary in Sleep rather than from the 1998 survey itself. But nothing in this evidence base measured next-day cognitive performance after one or two drinks in working adults. Everything known about alcohol and next-day thinking comes from hangover research at doses above one gram per kilogram — Gunn’s 2018 systematic review covered nineteen studies and 1,163 participants, of which eleven had enough data to pool; the meta-analysis found. Even there the picture is partial: working memory results were mixed, and divided attention and vigilance showed no clear effect, so the authors’ own finding is that specific components of cognition take the hit rather than cognition in general. Devenney’s 2019 participants were recruited in a pub, and one of that paper’s authors has disclosed alcohol-industry consultancy elsewhere; the disclosure statement for this particular study was not retrieved. So this piece is about your night and about one physiological system. It is not about your Tuesday afternoon, and anyone selling you a nightcap-dose cognitive claim is extrapolating across a gap nobody has measured.

Where the sedative effect actually lives

I went looking for the dose at which alcohol starts making you fall asleep faster. I expected a judgment call. What I found was a ladder.

That is the meta-analysis by Gardiner and colleagues in Sleep Medicine Reviews (2025;80:102030, online November 2024) — 27 studies, PRISMA, protocol pre-registered on OSF. I saw its abstract and publisher-page text, and its Results section through search-engine snippets. I have not opened the whole paper, and its funding and conflicts statement was never retrieved, which I would rather tell you than let silence imply it was clean.

The authors also report that objective blood alcohol values were absent across the included studies, and that 21 of the 27 were judged at risk of bias. The doses in that table are administered doses, not verified ones. Hold the whole ladder a little loosely for that reason.

Read it downward and the nightcap stops making arithmetic sense. Every cost on that ladder appears at a lower dose than the thing people drink for. REM onset is pushed back at roughly two drinks. REM duration starts shrinking around two or three. Falling asleep faster — the entire point of a nightcap — does not show up until about five.

Check the conversion yourself, because it is where this whole topic goes wrong. A US standard drink is 14 g of pure ethanol. Multiply 0.85 g/kg by an 80-kilogram body: 68 g, about five drinks. Do the same for 0.50 g/kg and you get 40 g, about three. The paper’s abstract calls 0.50 g/kg approximately two standard drinks; its Results section calls the same dose approximately three. Run the numbers and you will see the two answers correspond to different assumed body weights. That is why the doses are stated in grams per kilogram, why the table above gives a range where the paper gives two figures, and why every drink equivalent here should be read as approximate. Our piece on tart cherry juice ran into the same arithmetic problem from the opposite direction: a research dose that merely sounds like a serving.

One cross-reference is worth making, carefully, because neither source makes it. Examine’s page on alcohol reports the US National Institute on Alcohol Abuse and Alcoholism definition of heavy drinking as five or more standard drinks in a single day for men and four or more for women. Gardiner’s threshold for a measurable speeding of sleep onset works out, on the arithmetic above, to about five drinks for an eighty-kilogram adult. Those two numbers were arrived at independently by two organisations answering unrelated questions, and putting them side by side is my doing, not theirs. But side by side is where they are: the dose at which the sedative effect becomes measurable is roughly the dose the American alcohol institute uses to define a heavy drinking day.

Then there is the part that gets skipped. Total sleep time, sleep efficiency and wake after sleep onset — the three things people actually complain about — were not significantly affected at any dose, and the authors stress the wide 95% prediction intervals behind that result. That is not a null. A null is a study that could have found something and did not. An estimate with prediction intervals that wide is a study that could not have told you either way, and reporting it as “alcohol has no effect on total sleep time” would be putting a confident sentence on top of an uncertain number.

Here is what a real null looks like on this exact question. Feige and colleagues (Alcoholism: Clinical and Experimental Research 2006;30(9):1527–1537) ran a crossover design with a week between conditions, dosing ten healthy adults — five men and five women — to a blood alcohol level of either 0.03% or 0.10% at bedtime, for three consecutive nights, with polysomnography and subjective questionnaires each night and two further nights afterwards to catch any withdrawal effect. At 0.03% — a level a single drink will put most adults near — no clear effects could be detected. Nothing. And at the higher level, the two discontinuation nights showed no statistically significant effect on any sleep parameter either. Ten people is small, and a small study failing to find something is weaker evidence than a large one failing to find it. But this is a properly instrumented experiment, dosed to the question in the headline, reporting an outcome it looked for and did not see. That is a different object from an interval too wide to read, and telling the two apart is most of the skill. We have written about how to read a null at length.

A second study makes the point from another angle. Miyata and colleagues (Internal Medicine 2004;43(8):679–684) put 13 healthy female students, average age 21, through polysomnography on three nights: no alcohol, then 0.28 g/kg, then 0.69 g/kg. The lower dose sits below every threshold on the ladder above. Across all 13 subjects, the abstract reports no significant differences in any sleep parameter between the baseline night and either alcohol night. All 13 were also assessed with an alcohol patch test, and the results are reported split by that classification; whether the split was planned in advance is not stated at abstract level. In the six women classed as sensitive, REM as a percentage of sleep fell from 18.3 ± 6.2% at baseline to 9.8 ± 5.1% and 11.0 ± 2.8% on the two alcohol nights, and REM latency was significantly prolonged. In the seven insensitive women, nothing moved at all.

The paper is titled “REM Sleep is Impaired by a Small Amount of Alcohol in Young Women Sensitive to Alcohol.” Read that against the whole-sample result and you have this publication’s recurring pattern inside one document: a finding that is real within a subgroup of six, carried in the title as though it were the study’s result. The qualifier is right there — in young women sensitive to alcohol — and it is the part that falls off in citation. It also cuts the reader’s way. If you are one of the people who flushes, this might be your paper. Thirteen women in 2004 is nowhere near enough for anyone to tell you either way.

What one drink does instead

The best-matched study in this literature dosed exactly one glass of wine.

De Zambotti and colleagues (Sleep 2021;44(1):zsaa135) recruited 26 healthy adults for three counterbalanced nights in a laboratory — 15 men averaging 42, 11 women averaging 50, all screened by clinical polysomnography for insomnia and sleep apnoea, none of them abstainers or heavy drinkers. The beverage was wine, 150 mL of roughly 12% per glass: 14 g of ethanol, one US standard drink exactly. The low-dose condition was one glass for women and two for men, giving pre-sleep breath alcohol of 0.010% and 0.024%. Placebo was dealcoholized wine. Every participant blew 0.000 on arrival. This is a single night per condition, with a lopsided sex split and attrition — 26 were recruited, 19 completed all three nights, and the analysis rests on 21 high-dose, 23 low-dose and 23 placebo nights, measured over the first six hours, with dosing fixed by sex rather than body weight — hold that in mind for everything below.

On the low-dose night, heart rate ran faster than placebo through the first four hours, which the authors quantify as about 4% faster on that night. Vagally mediated heart-rate variability was suppressed — in men through hours one to four. Cardiac sympathetic activity was elevated, in men from hour three onward. Baroreflex sensitivity was suppressed. And the suppression of heart-rate variability did not significantly differ between the low dose and the high dose of three or four glasses — across 21 high-dose and 23 low-dose nights, one per person per condition, which is enough to say the extra drinks bought no visible extra suppression, and not enough to say the response has no gradient at all.

Meanwhile the sleep stages barely moved. There was no alcohol effect on REM or N2 percentage at either dose. The authors’ own conclusion is that the cardiovascular impact was, at these doses, “largely independent of alterations in sleep macrostructure.”

That study was funded by an NIAAA grant. Its authors also declare research funding unrelated to this work from Ebb Therapeutics, Fitbit, International Flavors & Fragrances and Noctrix Health. I believe this result — it is the cleanest thing in the register — and the Fitbit line still belongs in the sentence, particularly since wearable-derived evidence is about to do a lot of work here.

Two large real-world datasets say the same thing at population scale, and both come with a commercial fingerprint that has to be stated. Pietilä and colleagues (JMIR Mental Health 2018;5(1):e23, doi 10.2196/mental.9519) measured beat-to-beat heart intervals in 4,098 Finnish employees, mean age 45. In their lowest dose band — where about nine days in ten involved a single drink — a heart-rate-variability-derived recovery measure fell by 9.3 percentage units, and heart rate and the LF/HF ratio moved in that band too. That study was supported by a national innovation agency, and two of its authors were employed by Firstbeat Technologies, which makes the sensor and owns the proprietary recovery algorithm being reported; a third worked at Nokia’s digital health lab. It also only looked at the first three hours of the night, and the doses were self-reported.

Grosicki and colleagues (PLOS Digital Health 2026;5(3):e0001284, doi 10.1371/journal.pdig.0001284) analysed 20,968 wearable users across 5.1 million person-days. One drink more than a person’s own average, versus one less, was associated with nocturnal resting heart rate about 2.8 bpm higher in women and 2.4 in men, and heart-rate variability 3.8 ms and 3.3 ms lower. That study was funded by WHOOP, seven of its authors are WHOOP employees, six hold stock options, and the authors themselves note in print that demonstrating wearables’ sensitivity to behaviours like drinking could have favourable commercial implications for the company. Stating that is not an accusation; the disclosure is unusually thorough and the analysis was pre-registered. But a within-person association is not an effect, and their own summary is that many of the estimated effects were small.

Small is the right word. Under three beats per minute is not a transformed morning, and nobody should sell it as one. What makes it interesting is not the size — it is the location. It shows up at a dose where the sleep-stage summary does not move at all. The sexes disagree across datasets, too: the wearable cohort found consistently larger effects in women, while the Finnish cohort found the suppression similar across genders. That is genuinely unresolved and I am not going to resolve it for you.

One more piece of honesty, and this is the rare case where the mechanism story gets better rather than worse. The account usually offered for alcohol’s sleep-promoting effect — that it raises extracellular adenosine, which then inhibits wake-promoting neurons — was worked out in Sprague-Dawley rats. The human half now exists: a 2018 PET study in 49 healthy volunteers, average age 26, found that an infusion of ethanol raised the availability of A1 adenosine receptors by as much as 26% across several brain regions. I have that at abstract level, and it is a receptor-availability finding rather than a sleep finding — that study’s cognitive testing was done during the acute exposure, not the next day. So the system is confirmed to respond to alcohol in humans, and the specific chain from a drink at nine to falling asleep faster at eleven is still rat data. Any sentence beginning “alcohol works on your sleep by” is standing on the rat half.

How the deep-sleep story got into everything you have read

If you have read anything about alcohol and sleep, you have read that it increases deep sleep early in the night and steals it back later. That claim has a traceable history, and the history is better than the claim.

2013. Ebrahim and colleagues published a qualitative review of 20 studies in Alcoholism: Clinical and Experimental Research (doi 10.1111/acer.12006). It reported reduced sleep onset latency at all dosages and concluded that alcohol clearly increases slow-wave sleep in the first part of the night at all doses, across sexes and ages. It became the source for most of what was written on this subject for the next decade. Its corresponding author is medical director of a commercial private sleep clinic on Harley Street — a fact about the paper, not a verdict on it.

2015. Eight sleep researchers wrote to the same journal (doi 10.1111/acer.12712) arguing the statistics could not carry the conclusions. Their specifics: 28 of the 38 subject groups in the review contained ten or fewer people. Non-significant differences had been reported as increases. And a measure called “% difference N3” had been constructed by subtracting or dividing group means while ignoring standard deviations entirely. Their worked example is easy to check: one included study reported slow-wave sleep going from 13.41 ± 7.01 to 12.72 ± 7.17 on the first alcohol night, p = 0.412 — a non-significant decrease — and the review recorded it as a 0.68% difference. Across the whole night, they counted 7 of 37 studies showing any statistically significant difference in deep sleep, and 4 of those 7 were significant decreases. In the first half of the night, 9 of 18 showed significant but small increases. Their prescription: someone should run a proper meta-analysis.

I read that letter in full. It is worth noting that forensic sleep medicine is heavily represented among its authors and that it closes on the alcohol-induced-sleepwalking legal defence, which is a stated professional interest of several of them. That may motivate the critique. It does not touch the arithmetic, which anyone can check. And the underlying complaint is not confined to eight people with a stake in it: writing in Sleep in 2024, Colrain and Baker describe the literature on alcohol’s acute effects on sleep architecture as “sparse and dominated by studies with small sample sizes.” That is the field describing itself, in an editorial, nine years later.

2015, three pages later. Ebrahim and colleagues replied in the same issue, under the title “Alcohol and Sleep Review: Sound Statistics and Valid Conclusions” (Alcohol Clin Exp Res 2015;39(5):944–946, doi 10.1111/acer.12708). I could not read it. Every access route I tried was paywalled, and the journal states on the page that no abstract was provided for the article. So you are reading one side of a two-sided disagreement, and I would rather say that plainly than pretend the reply does not exist or characterise an argument I have not seen. Nobody fabricated anything here. Nobody hid anything. The disagreement is a public document in two parts, and one part costs money.

2024. Someone ran the meta-analysis. It is the one at the top of this article, and the low-dose onset benefit does not survive it.

The consequence. None of this reached the reader. A sleep calculator at thesleepmath.com/alcohol-sleep-calculator cites Ebrahim et al. (2013) as “a meta-analysis of 27 studies.” Ebrahim is a qualitative review of 20 studies. The 27-study meta-analysis is Gardiner. The paper that was challenged and the paper that answered the challenge have been fused into a single citation, and the older one has been promoted to a study design it never claimed to be.

The three-hour rule, and what is underneath it

Examine.com, the most rigorous evidence aggregator in consumer health, has no entry connecting alcohol and sleep. Its alcohol page covers formulation, dose thresholds and cancer risk; sleep does not appear among the effects. Its sleep category page lists the supplements of most interest for sleep — apigenin, cannabidiol, hops, magnesium, melatonin, valerian and others — and alcohol is not among the factors discussed at all. Alcohol is also absent from the American Academy of Sleep Medicine’s April 2026 announcement of its guideline on combination therapy for chronic insomnia, which I will come back to.

So the careful sources treat a sleep aid that thirteen per cent of adults report using as a topic that does not exist. Into that vacuum goes the rule everybody repeats. The Sleep Foundation advises avoiding alcohol within three to four hours of bedtime, with no primary citation attached to the interval. Rules of this shape are something of a house speciality — we have taken apart the same problem in melatonin dosing. Gardiner’s paper states that there were no data on a fixed dose consumed at varying times before sleep.

Then there is the one controlled experiment that tested a longer interval than either rule requires. Landolt and colleagues (J Clin Psychopharmacol 1996;16(6):428–436) gave 0.55 g/kg of ethanol six hours before bedtime to ten healthy men averaging 61 years old. Breath ethanol had declined to zero in every subject before the sleep episode began. Sleep was still perceived as more superficial than on the mineral-water control. Sleep efficiency, total sleep time, stage 1 and REM sleep were all reduced, and wakefulness in the second half of the night doubled. Ten men, thirty years ago, two decades older than you, read at abstract level — all true, and it is still a controlled experiment in which the alcohol was entirely gone and the back half of the night fell apart anyway. Its daytime performance measures were taken; I could not retrieve the result.

The clearance arithmetic that underwrites “wait until it’s out of your system” is shakier than it looks in any case. Forensic pharmacology puts elimination at roughly 10–15 mg of ethanol per 100 mL of blood each hour after drinking on an empty stomach — call it about one standard drink an hour for an average adult — and 15–20 mg per 100 mL each hour when fed, with about a threefold spread between individuals. A nightcap is almost always taken fed, so the popular one-drink-per-hour rule is, unusually, conservative for the actual use case. It is also a band, not a number, and it is answering a question that Landolt’s data suggest is the wrong one. This is a clock rule with the same structural problem as the caffeine cutoff: it looks like it is measuring the intervention and is actually measuring something adjacent.

What it costs when it isn’t one drink

Breathing. A 2018 meta-analysis of 14 randomised crossover trials, 422 participants, 72% male (doi 10.1016/j.smrv.2018.05.007) found the apnoea–hypopnoea index rose by 2.33 events per hour (95% CI 1.41–3.25, I² = 62%) and mean oxygen saturation fell by 0.60 (95% CI −0.72 to −0.49). The effect was concentrated in people with existing obstructive sleep apnoea and people with a snoring history. What that design can show is a pooled within-person shift; what it cannot tell you is whether 2.33 events matters clinically for any individual, or whether it generalises past a heavily male sample. And the composite hides a null of the honest kind: taken separately, the apnoea index alone (+0.60, 95% CI −1.54 to 2.75) and the hypopnea index alone (+1.30, 95% CI −0.06 to 2.67) each failed to reach significance. The combined measure moved; neither component did on its own.

There is also a counter-finding, in the subgroup most at risk. An open-label study of 26 stable COPD inpatients, median age 65, given 0.5 g/kg immediately before lights out (PMCID PMC4330402), found REM as a proportion of sleep fell by 3.1% (95% CI 0.2–6.0, p = 0.020) — but the six subjects with an apnoea–hypopnoea index of 15 or more had fewer events on alcohol, a mean reduction of 11 (95% CI 1–20, p = 0.046). Wrong population, small, open-label, read at abstract level. It still points the other way, and burying it would be the exact failure this site exists to point at in other people.

The second half of the night. McCullar and colleagues (Sleep 2024;47(4):zsae003) ran three consecutive laboratory nights and used additive modelling to locate the damage rather than average over it. Excess wakefulness appeared only from about five and a half hours after sleep onset to the end of the night, and only on the first drinking night. Deep sleep was elevated from roughly 1.8 to 6.5 hours in. Total REM fell about 11 minutes on the first alcohol night and around 4 minutes on the second and third. Whether that attenuation is genuine tolerance is disputed: the study’s own reading and the accompanying journal commentary describe the habituation differently, and the two use different measures. That commentary is by Colrain and Baker — who are also two of the authors of the wine study earlier in this article, which is worth knowing when they are the ones telling you how to read someone else’s data. Unresolved either way.

Cardiovascular, at a real evening’s dose. A 2025 prospective study of 40 healthy adults, mean age 30, 63% female (Nutrients 2025;17(9):1470, doi 10.3390/nu17091470) ran three alcohol-free baseline days, three evenings at 40 g for women and 60 g for men, and three days after. Nocturnal resting heart rate went 63.6 → 66.6 → 64.9 bpm (p < 0.001) — up about three beats, then back. There was no significant change in objective sleep architecture and none in daytime activity. Subjective sleep quality fell anyway. Forty people, no funding statement retrieved. The authors then take their three beats, borrow a mortality gradient from a chronic-exposure epidemiological study, and derive a predicted percentage increase in mortality. That step maps a long-run association onto a transient overnight change, in a study with no mortality endpoint and no follow-up. It is the same error class as the sleep calculator above, committed by researchers rather than a website, and I am not carrying the number across.

What is not here. No safety data specific to a single drink. Nothing on habitual light evening drinking over months or years. No laboratory study has followed anyone past three consecutive nights.

What to change, in order

Three levers, ranked by the size of the evidence behind them. All three come from the same observational wearable dataset unless stated, which means they are within-person associations, self-reported exposure, one device — I will say that once and not repeat it.

1. Protect the sleep opportunity. Do not shorten the night. This is the largest effect anywhere in this evidence base. After five drinks above a person’s average, the highest-sleep third of nights differed from the lowest by 2.9 bpm of resting heart rate and 7.2 ms of heart-rate variability, with effect sizes of 0.65 and 0.58. Sleeping less than usual after drinking amplified the disruption; for anyone who drinks in the evening and keeps a fixed alarm, the two insults stack. If you know in advance you will be drinking, the thing to move is the alarm, not the glass.

2. Do not train it off. Counterintuitively, more physical activity than usual on a drinking day amplified the overnight cardiovascular disruption rather than offsetting it. Low-activity days at five drinks came out 0.7 bpm lower in heart rate and 2.1 ms higher in variability than high-activity days, and yielded a few more minutes of sleep. The hard morning session is moving the wrong lever.

3. Move the drink earlier — but expect less, and know what contradicts it. Drinking 60 minutes earlier than usual, rather than 60 later, was associated with 0.87 bpm lower resting heart rate and 1.5 ms higher variability in women; in 20-to-29-year-olds, 1.2 bpm and 3.7 ms. That is somewhere between a fifth and a third of the size of lever one, depending on which measure you use. It also came with slightly shorter sleep in both sexes, which is a real cost against lever one. And Landolt’s men had zero breath alcohol at lights-out and lost REM anyway. Take the earlier drink if it is free; do not treat it as protection.

If the drink is doing a job

This last part is not a lever. It is an observation, and it is for a narrower reader: the one who drinks because they cannot switch off.

In everything I could find, exactly one study looked at that behaviour directly. Roehrs and colleagues (Neuropsychopharmacology 1999;20(3):279–286) took 20 moderate social drinkers aged 21 to 45 — eleven with polysomnography-documented insomnia, nine sleeping normally — gave them 0.5 g/kg or placebo on sampling nights, then let them choose for themselves. The insomniacs chose significantly more ethanol refills than the normal sleepers, averaging 0.45 g/kg a night; the normal sleepers took significantly more placebo. The authors conclude that the acute sleep and mood effects are bound up with alcohol’s reinforcing properties as a hypnotic in that group. I have that at abstract level, and the individual sleep outcomes were not retrievable there. But run 0.45 g/kg against an 80-kilogram body and you get about 36 g — roughly two and a half drinks. People who use alcohol for sleep converge, unprompted, on a dose sitting inside the REM-cost band and well short of the sedative one.

Which is where the guideline I mentioned earlier comes in. In April 2026 the American Academy of Sleep Medicine issued a clinical practice guideline on combination therapy for chronic insomnia — a condition it puts at roughly 10 to 15% of adults. That scope matters: the question it set out to answer was whether pairing medication with cognitive behavioural therapy beats either alone, which its 2017 and 2021 guidelines had not evaluated. Its two recommendations are both conditional and both rest on low certainty of evidence. The first suggests combination therapy over medication alone. The second suggests against combination therapy rather than CBT-I by itself, on the grounds that the behavioural treatment alone often produces durable improvement without the added risks of pharmacotherapy. The guideline’s lead author is quoted in the announcement saying that “CBT-I by itself is the most efficacious first-line treatment for insomnia.” I have read the announcement, not the guideline; its task force includes members with pharmaceutical and digital-health consultancies, two AASM employees, and one member of the AASM board, all disclosed. The point is not that you should stop drinking. It is that if the drink is treating something, there is a treatment for that thing with actual evidence behind it, and a depressant that costs REM from around two drinks is not competing on equal terms.

Which brings it back to the screen you woke up to. The reason none of this appeared in your sleep-stage summary is that the sleep-stage summary was never the instrument measuring it. Two systems come apart on a drinking night, and only one of them is made of stages. Whether the band on your wrist measures the other one accurately is a separate question, and one this evidence cannot answer for you.

About this article

Written by Drew Anton. Drew covers behaviour, sleep, and the compounds people take for both — doses, timing, and what the label does and does not tell you. Not a physician or research scientist — reads the primary literature closely and refuses to round up.

Medical review: None. NeuriFuel does not currently have a licensed clinician on the editorial team, and this article has not been medically reviewed. We state this rather than implying an authority we do not have. See our About page for our full methodology.

Sources: 25 sources — three meta-analyses, one qualitative review together with its published critique and the reply to that critique, six controlled human sleep-laboratory studies, one open-label inpatient study, three observational cohorts, one naturalistic cognitive study, one epidemiological survey, one human neuroimaging and performance study, two forensic-pharmacology reviews, one journal commentary, one clinical-guideline announcement, one consumer evidence aggregator, and one animal-model book chapter. Six were read in full text: Pressman 2015, Colrain & Baker 2024, de Zambotti 2021, Grosicki 2026, and the Examine and AASM pages as supplied. The rest were read at abstract or index-record level, and this article’s verbs reflect that throughout — where a paper’s result could not be retrieved, it is not described. Funding and conflict disclosures could not be retrieved for the majority of the studies cited here, including the 2024 meta-analysis that carries the dose thresholds; that is marked rather than left silent, and where disclosures were retrieved they are flagged inline next to the study they affect, including for studies supporting this article’s conclusion. Values that no source specified are marked “Data not specified” rather than filled in.

Corrections: Found an error? Write to hello@neurifuel.com with a source and we will fix it and log the correction.

Last updated: 23August 2026

References

1. Gardiner C, Weakley J, Burke LM, Roach GD, Sargent C, Maniar N, Huynh M, Miller DJ, Townshend A, Halson SL. The effect of alcohol on subsequent sleep in healthy adults: a systematic review and meta-analysis. Sleep Medicine Reviews 2025;80:102030. doi 10.1016/j.smrv.2024.102030 · PMID 39631226. Read at abstract, publisher page and Results-snippet level. Funding and conflicts: not retrieved.
2. Pressman MR, Grunstein RR, Mahowald MW, Schenck CH, Montplaisir JY, Cramer Bornemann M, Zadra A, Buchanan PR. Alcohol and sleep review: flawed design, methods, and statistics cannot support conclusions. Alcoholism: Clinical and Experimental Research 2015;39(5):941–943. doi 10.1111/acer.12712. Read in full. No funding or disclosure statement appears in the letter.
3. Ebrahim IO, Shapiro CM, Williams AJ, Fenwick PB. Alcohol and sleep I: effects on normal sleep. Alcoholism: Clinical and Experimental Research 2013;37(4):539–549. doi 10.1111/acer.12006. Corresponding author is medical director of a commercial private sleep clinic. Funding: Data not specified.
4. Ebrahim IO, Fenwick P, Williams AJ, Shapiro CM. Alcohol and sleep review: sound statistics and valid conclusions. Alcoholism: Clinical and Experimental Research 2015;39(5):944–946. doi 10.1111/acer.12708 · PMID 25870937. Content not retrieved — cited for its existence and title only.
5. de Zambotti M, Forouzanfar M, Javitz H, Goldstone A, Claudatos S, Alschuler V, Baker FC, Colrain IM. Impact of evening alcohol consumption on nocturnal autonomic and cardiovascular function in adult men and women: a dose–response laboratory investigation. Sleep 2021;44(1):zsaa135. doi 10.1093/sleep/zsaa135 · PMID 32663278. Read in full. NIAAA grant R21-AA024841; authors declare unrelated research funding from Ebb Therapeutics, Fitbit, International Flavors & Fragrances and Noctrix Health.
6. Pietilä J, Helander E, Korhonen I, Myllymäki T, Kujala UM, Lindholm H. Acute effect of alcohol intake on cardiovascular autonomic regulation during the first hours of sleep in a large real-world sample of Finnish employees. JMIR Mental Health 2018;5(1):e23. doi 10.2196/mental.9519 · PMID 29549064. Industry-linked: funded by TEKES; two authors employed by Firstbeat Technologies, manufacturer of the sensor and owner of the recovery algorithm reported; one author employed by Nokia Technologies’ Digital Health Lab.
7. Grosicki GJ, Robinson AT, Joyner MJ, Carter JR, von Hippel W, Presby DM, Fielding F, Bigalke JA, Kim J, Chapman C, Holmes KE. Real-world effects of alcohol on heart rate, sleep, and physical activity by age and sex. PLOS Digital Health 2026;5(3):e0001284. doi 10.1371/journal.pdig.0001284. Read in full. Industry-funded: funded by WHOOP, Inc.; seven authors are WHOOP employees, six hold stock options.
8. Landolt HP, Roth C, Dijk DJ, Borbély AA. Late-afternoon ethanol intake affects nocturnal sleep and the sleep EEG in middle-aged men. Journal of Clinical Psychopharmacology 1996;16(6):428–436. doi 10.1097/00004714-199612000-00004 · PMID 8959467. Abstract level. Funding: Data not specified.
9. Kolla BP, Foroughi M, Saeidifard F, Chakravorty S, Wang Z, Mansukhani MP. The impact of alcohol on breathing parameters during sleep: a systematic review and meta-analysis. Sleep Medicine Reviews 2018;42:59–67. doi 10.1016/j.smrv.2018.05.007 · PMID 30017492. Abstract and excerpt level. Funding: Data not specified.
10. Holmedahl NH, Øverland B, Fondenes O, Ellingsen I, Hardie JA. Alcohol at bedtime induces minor changes in sleep stages and blood gases in stable chronic obstructive pulmonary disease. Sleep & Breathing 2016. PMCID PMC4330402; DOI not retrieved. Abstract level. Open-label design. Funding: Data not specified.
11. McCullar KS, Barker DH, McGeary JE, Saletin JM, Gredvig-Ardito C, Swift RM, Carskadon MA. Altered sleep architecture following consecutive nights of presleep alcohol. Sleep 2024;47(4):zsae003. doi 10.1093/sleep/zsae003. Abstract level. Funding: Data not specified.
12. Colrain IM, Baker FC. To drink perchance to sleep: a commentary on “Altered sleep architecture following consecutive nights of pre-sleep alcohol” by McCullar et al. Sleep 2024;47(4):zsad333. doi 10.1093/sleep/zsad333 · PMID 38205873. Read in full. Supported by NIAAA grant R21AA028617, with no financial or non-financial disclosures declared. Both authors are also authors of reference 5.
13. Strüven A, Schlichtiger J, Hoppe JM, Thiessen I, Brunner S, Stremmel C. The impact of alcohol on sleep physiology: a prospective observational study on nocturnal resting heart rate using smartwatch technology. Nutrients 2025;17(9):1470. doi 10.3390/nu17091470 · PMCID PMC12073130. Abstract and excerpt level. Funding: Data not specified.
14. Feige B, Gann H, Brueck R, Hornyak M, Litsch S, Hohagen F, Riemann D. Effects of alcohol on polysomnographically recorded sleep in healthy subjects. Alcoholism: Clinical and Experimental Research 2006;30(9):1527–1537. doi 10.1111/j.1530-0277.2006.00184.x · PMID 16930215. Abstract level. Funding: Data not specified.
15. Gunn C, Mackus M, Griffin C, Munafò MR, Adams S. A systematic review of the next-day effects of heavy alcohol consumption on cognitive performance. Addiction 2018;113(12):2182–2193. doi 10.1111/add.14404 · PMCID PMC6282576. Abstract level. Funding: Data not specified.
16. Devenney LE, Coyle KB, Verster JC. Cognitive performance and mood after a normal night of drinking: a naturalistic alcohol hangover study in a non-student sample. Addictive Behaviors Reports 2019. doi 10.1016/j.abrep.2019.100197 · PMCID PMC6597921. Abstract level. Funding: Data not specified. One author has disclosed alcohol-industry consultancy in other publications; the disclosure statement for this paper was not retrieved.
17. Jones AW. Evidence-based survey of the elimination rates of ethanol from blood with applications in forensic casework. Forensic Science International 2010. PMID 20304569. Abstract level. Volume and pages not retrieved. Funding: Data not specified.
18. Jones AW. Pharmacokinetics of ethanol — issues of forensic importance. Forensic Science Review 2011;23(2):91–136. PMID 26231237. Abstract and PDF-excerpt level. Funding: Data not specified.
19. Johnson EO, Roehrs T, Roth T, Breslau N. Epidemiology of alcohol and medication as aids to sleep in early adulthood. Sleep 1998;21:178–186. doi 10.1093/sleep/21.2.178 · PMID 9542801. Cited via reference 12; not opened.
20. American Academy of Sleep Medicine. Announcement of a clinical practice guideline on combination therapy for chronic insomnia, Journal of Clinical Sleep Medicine, 13 April 2026. DOI, volume and pages: Data not specified. Press release read in full; guideline not read. Task force disclosures include pharmaceutical and digital-health consultancies, research support from Sleep Number, two AASM employees and one AASM board member.
21. Examine.com — alcohol page and sleep category page, including the page’s report of the US National Institute on Alcohol Abuse and Alcoholism heavy-drinking definition. Secondary evidence aggregator, read in full as supplied. Examine states it takes no industry funding or advertising; not independently verified.
22. Thakkar MM. Alcohol-induced sleep promotion: adenosine and orexin mechanisms. Springer. doi 10.1007/978-3-319-23078-8_9. Rat model (Sprague-Dawley). Index-record level. Funding: Data not specified.
23. Miyata S, Noda A, Ito N, Atarashi M, Yasuma F, Morita S, Koike Y. REM sleep is impaired by a small amount of alcohol in young women sensitive to alcohol. Internal Medicine 2004;43(8):679–684. doi 10.2169/internalmedicine.43.679 · PMID 15468965. Abstract level. Funding: Data not specified.
24. Elmenhorst E-M, Elmenhorst D, Benderoth S, Kroll T, Bauer A, Aeschbach D. Cognitive impairments by alcohol and sleep deprivation indicate trait characteristics and a potential role for adenosine A1 receptors. PNAS 2018;115(31):8009–8014. doi 10.1073/pnas.1803770115 · PMID 30012607. Abstract level. Funding: Data not specified.
25. Roehrs T, Papineau K, Rosenthal L, Roth T. Ethanol as a hypnotic in insomniacs: self administration and effects on sleep and mood. Neuropsychopharmacology 1999;20(3):279–286. doi 10.1016/S0893-133X(98)00068-2 · PMID 10063488. Abstract level. Supported by US Public Health Service research funding; no conflict statement retrieved.