Alcohol and sleep

Key findings

1. 1
   Strong evidenceMeta-analysis · Systematic review · Lab study

   Alcohol in the evening delays and reduces REM (dream) sleep, and the more you drink, the bigger the effect.

   Meta-analysis of 27 studies in healthy adults: REM onset delayed 18.0 min (95% CI 8.8–27.1) and REM duration reduced 11.3 min (95% CI −20.8 to −1.8), or 2.8 percentage points of the night. REM loss grew ~40 min per 1 g/kg. Significant from ~0.35 g/kg (REM latency) and ~0.50 g/kg (REM duration), roughly 1.5–2 US standard drinks for a 62-kg adult.^1,2

   Limits Nearly all studies were in young, healthy, light-to-moderate drinkers dosed within ~3 h of bed. Time in bed was often uncontrolled, and the g/kg-to-drinks conversion is approximate.

2. 2
   Moderate evidenceMeta-analysis · Lab study

   A drink or two barely speeds up falling asleep; a clear effect on sleep onset shows up only at high doses, when the downsides are larger.

   Pooled sleep-onset latency was not significantly changed (−2.0 min, 95% CI −4.3 to 0.2). It shortened 6.4 min per 1 g/kg and became significant only around 0.85 g/kg (~4 US or ~5 Australian standard drinks). In 18–21-year-olds a dose reaching BAC 0.084% did not shorten onset.^1–3

   Limits Most control subjects already fell asleep in under 20 min, which limits the room for improvement. People with genuinely slow onset may feel a larger sedative effect, but tolerance develops quickly.

3. 3
   Moderate evidenceLab study · Randomised trial · Review

   As alcohol wears off during the night, the second half of sleep tends to become lighter and more broken.

   Lab studies show more slow-wave sleep and fewer wakings early, then more wake and stage-1 sleep later. In 93 young adults dosed to a breath alcohol level of ~0.11%, consolidation rose in the first half and fell in the second, and morning sleep quality was rated lower. At a bedtime BAC of 0.10%, rebound signs appeared in the second half.^1–5

   Limits The 2025 meta-analysis could not confirm a pooled whole-night increase in wake after sleep onset or a loss of total sleep, because too few studies reported half-night data. The pattern is clearest at moderate-to-high doses.

4. 4
   Moderate evidenceReview · Meta-analysis · Lab study

   Alcohol can add some deep sleep early in the night, but this is offset by lighter sleep later.

   Most studies find more slow-wave sleep in the first half of the night, with total-night slow-wave sleep raised mainly at high doses. In pooled data, whole-night N3 duration and proportion were not significantly changed.^1,2,5

   Limits First-half deep-sleep gains are dose-dependent and transient, and they do not mean better sleep overall.

5. 5
   Mixed evidenceLab study · Meta-analysis

   The idea that REM sleep simply 'catches up' later in the night is not reliable; overall REM is usually still reduced.

   Some studies report more REM in the second half. Others, such as a study of 24 people aged 18–21 at BAC ~0.08%, found no rebound, and the meta-analysis found total REM reduced.^1–3

   Limits Rebound may depend on dose, age and how much REM was lost early.

6. 6
   Mixed evidenceLab study · Cohort

   Even a single drink can change night-time heart rhythm, although its effect on sleep stages is small.

   At a bedtime blood alcohol of 0.03%, polysomnography showed no clear changes. In 4,098 employees, however, low intake (≤0.25 g/kg, average ~1 drink of 12 g) lowered the HRV-based recovery index by 9.3 percentage units and raised sleeping heart rate by ~1.4 bpm.^5,6

   Limits The HRV recovery index is a proprietary metric. Whether a ~1-bpm change matters for how rested people feel is unknown.

7. 7
   Moderate evidenceCohort · Lab study

   Evening alcohol raises your heart rate and lowers heart-rate variability during sleep, step by step with the amount you drink.

   Real-world data (n=4,098): recovery index fell 9.3, 24.0 and 39.2 percentage units, and sleeping HR rose 1.4, 4.0 and 8.7 bpm, after low, moderate and high intake (≤0.25, 0.25–0.75, >0.75 g/kg). In the lab (n=26), 1–2 and 3–4 drinks raised nocturnal HR, suppressed vagal HRV and baroreflex sensitivity, and raised sympathetic activity.^6–8

   Limits The large study was observational and had no record of drinking time; some authors were employed by the wearable vendor. The lab studies were small.

8. 8
   Limited evidenceMeta-analysis · Expert consensus

   How long to leave between your last drink and bed has barely been tested; three to four hours is common expert advice rather than a proven cutoff.

   In the 2025 meta-analysis, alcohol was given within ~3 h of bed in almost all studies, so the effect of timing could not be established. AASM public advice is to allow 'three to four hours between drinking and bedtime'.^1,9

   Limits Absence of timing data is not evidence that timing does not matter. The pharmacokinetics suggest that a larger dose needs a longer gap.

9. 9
   Limited evidenceLab study

   A moderate amount of alcohol in the late afternoon can still disturb that night's sleep, even if you feel sober by bedtime.

   In 10 men aged ~62, 0.55 g/kg taken 6 h before bed (zero breath alcohol at lights-out) reduced sleep efficiency, total sleep time and REM, and doubled wakefulness in the second half of the night.^10,11

   Limits A single small study in older men. A forced-desynchrony study in young adults at a lower dose found few changes.

10. 10
    Strong evidenceReview · Guideline

    The body clears alcohol at a roughly fixed pace: on average about half of a US standard drink per hour, and nothing speeds it up.

    The average elimination capacity is ~7 g/h for a 70-kg adult (often rounded to 'one drink per hour'), with 3–4-fold differences between people. Blood alcohol typically falls 10–20 mg/100 mL per hour, with 15 mg/100 mL/h a good average for moderate drinkers. A US standard drink is 14 g, so an average adult needs ~2 h to clear one.^12–14

    Limits The 2-h-per-US-drink figure is a derived average. Smaller people, fasting drinkers and some genotypes differ, and heavy drinkers can clear faster through enzyme induction.

11. 11
    Limited evidenceReview · Lab study

    Three hours is usually enough to clear one drink before bed, but two or more drinks need more time.

    Derived estimate: at ~7 g/h, one US drink (14 g) clears in ~2 h and two drinks (28 g) in ~4 h, plus absorption time. In a study of older men, a dose of ~3 US drinks (0.55 g/kg) still disturbed sleep when taken 6 h before bed.^10,12,13

    Limits This is a pharmacokinetic calculation, not a sleep trial. Sleep effects may outlast measurable blood alcohol.

12. 12
    Moderate evidenceMeta-analysis · Lab study · Randomised trial

    Alcohol relaxes the throat muscles and can make snoring and breathing pauses during sleep worse, especially if you already snore.

    Meta-analysis of 14 experimental studies (n=422): AHI +2.33 events/h (95% CI 1.41–3.25), +4.20 in snorers and +7.10 in people with OSA; mean SpO2 −0.60 and nadir SpO2 −1.25 points. Across 21 epidemiological studies, higher intake was linked to a 25% higher risk of sleep apnea (RR 1.25, 95% CI 1.13–1.38).^15–18

    Limits The epidemiological meta-analysis had high heterogeneity (I² 82%) and publication bias. Loud snoring with gasping warrants a clinician's assessment; the app should not diagnose.

13. 13
    Moderate evidenceCross-sectional · Randomised trial · Lab study

    People who drink more tend to snore more, and in experiments a drink before bed increased snoring-related breathing events.

    In 11,905 adults, heavy drinkers had higher odds of snoring (OR 1.38, 95% CI 1.22–1.57). In 20 asymptomatic men, alcohol raised apneas from 20 to 110 and desaturations from 118 to 226, and desaturations stayed raised the next, alcohol-free night. In two benign snorers, alcohol induced obstructive apnea in the first hour of sleep.^17–19

    Limits The cohort data are cross-sectional. The experimental studies are small and old.

14. 14
    Moderate evidenceRandomised trial · Lab study

    Using alcohol to get to sleep stops working within about a week, and people then tend to drink more for the same effect.

    In people with insomnia, 0.6 g/kg before bed increased total sleep and deep sleep on night 2, but these gains were gone by night 6. Six nights of pre-treatment led to more self-chosen refills. In a blind choice, insomniacs picked alcohol on 67% of nights compared with 22% for normal sleepers.^20–22

    Limits Small lab samples (n=20–24 per experiment). Long-term escalation to dependence was not directly measured.

15. 15
    Limited evidenceCross-sectional

    Using alcohol as a sleep aid is common, which is part of why its downsides are worth knowing.

    In a representative 1996 survey of 2,181 US adults aged 18–45, 13% had used alcohol to help them sleep in the past year, and 15% of those had done so for more than a month. Difficulty falling asleep was the strongest predictor.²³

    Limits One old regional survey; current prevalence may differ.

16. 16
    Moderate evidenceCohort

    Regular or heavy drinking is linked to more trouble staying asleep later on, and poor sleep is linked to drinking more.

    Helsinki Health Study (n≈7,300, 5–7 y): heavy drinking predicted later insomnia symptoms (OR 1.48, 95% CI 1.22–1.80), binge drinking OR 1.26, and insomnia predicted later heavy drinking (OR 1.34). Japanese civil servants (n=2,861, 5 y): drinking on 3+ days a week was linked to difficulty staying asleep (OR 1.48, 95% CI 1.16–1.90), while daily drinkers reported less difficulty falling asleep (OR 0.42).^24,25

    Limits Observational, with self-reported drinking and sleep, so residual confounding (stress, mood) is likely.

17. 17
    Strong evidenceReview · Expert consensus

    In people who drink heavily over a long time, sleep problems are very common and can persist for months after stopping, so this is worth discussing with a clinician.

    Reviews describe insomnia as common in alcohol use disorder, bidirectional with drinking, persisting into abstinence and predicting relapse. CBT for insomnia is a promising treatment.^26–28

    Limits This applies to alcohol use disorder, not occasional drinkers. The app should not screen for or diagnose it.

18. 18
    Moderate evidenceLab study · Cohort

    Evening drinking is one of several common, changeable reasons for waking in the night or too early in the morning.

    Lab studies show more wake in the later part of the night after alcohol (doubled second-half wake in one study), and cohort data link frequent drinking to trouble staying asleep.^4,10,25

    Limits No study ranks alcohol against other causes (noise, light, heat, stress, breathing problems, age, mood) of night waking in the general population.

19. 19
    Mixed evidenceLab study · Meta-analysis · Cohort · Guideline

    Women may be more affected than men by the same amount of alcohol, partly because the same drinks produce a higher blood alcohol level.

    At the same peak breath alcohol (~0.11%), women's sleep time, efficiency and night waking were more disrupted than men's (n=93). Men and women showed different nocturnal heart-rate patterns, and morning blood pressure rose only in women after high doses. The meta-analysis found no sex effect, but data were limited, and a large real-world study found similar HRV effects in both sexes.^{1,4,6,7,29,30}

    Limits Few studies report results by sex, and hormonal phase is rarely controlled.

20. 20
    Limited evidenceLab study

    A moderate drink about an hour before bed slightly lowers the body's melatonin that night.

    In 29 young adults, ~0.5 g/kg finished 1 h before bed lowered salivary melatonin by 15% at 140 min and 19% at 190 min compared with placebo.³¹

    Limits Single small study. The effect on circadian timing was not measured.

21. 21
    Limited evidenceCohort · Meta-analysis · Review

    Caffeine does not cancel out alcohol: it does not clear alcohol faster, and the two disturb sleep in different ways.

    In a 6-week real-world study of financial traders, alcohol lowered subjective sleep quality and caffeine reduced sleep quantity; together each partly masked the other. A meta-analysis found caffeine did not significantly mask subjective intoxication. Alcohol elimination runs at a near-fixed hepatic rate.^12,32,33

    Limits Small real-world cohort. Mixed caffeinated-alcohol drinks raise separate safety concerns not covered here.

22. 22
    Strong evidenceGuideline

    Never combine alcohol with sleeping pills or sleep-aid antihistamines; check with a pharmacist or doctor about any sleep medicine you take.

    The zolpidem label states 'Do not take AMBIEN if you drank alcohol that evening or before bed' and lists additive CNS depression and a higher risk of complex sleep behaviours such as sleep-driving. NIAAA lists drowsiness, slowed or difficult breathing, impaired motor control and memory problems for alcohol with zolpidem, eszopiclone, temazepam or estazolam, and increased drowsiness with OTC diphenhydramine or doxylamine and herbal sedatives.^34–36

    Limits This is a safety rule, not a sleep-quality claim. NIAAA's list does not cover melatonin, and no specific interaction is claimed here.

23. 23
    Strong evidenceGuideline

    A 'drink' means about 14 g of alcohol in the US: roughly a 12-oz beer at 5%, a 5-oz glass of wine at 12%, or a 1.5-oz shot of spirits.

    NIAAA defines a US standard drink as ~14 g (0.6 fl oz) of pure alcohol. Other countries use smaller units (e.g. Australia 10 g, as in the 2025 meta-analysis).^1,14

    Limits Real pours are often larger than a standard drink.

24. 24
    Moderate evidenceGuideline · Review

    Cutting back on evening alcohol is a sensible habit, but on its own it is not a treatment for long-term insomnia.

    The AASM 2021 guideline suggests not using sleep hygiene, which includes avoiding alcohol, as a single-component therapy for chronic insomnia (conditional). A review found support for each hygiene item but few direct tests in the general population.^37,38

    Limits This is about chronic insomnia disorder. Evening alcohol still matters for nightly sleep quality.