It changes your sleep more than it ruins it: across 27 sleep-lab studies in healthy adults, alcohol cut REM sleep by about 11 minutes a night and pushed it later, even at low doses, while its effects on total sleep were uncertain3. The belief that a drink helps you sleep held only partly, and only at high doses.
Here is what was measured, what changed, and how big the changes were.
What was studied
The main evidence is a 2025 systematic review and meta-analysis, a study that pools the results of many trials into one estimate3. It included 27 controlled studies in healthy adults aged 18 to 70. Each gave a measured dose of alcohol and then recorded a full night of sleep with polysomnography: sensors on the head and face that read brain, eye and muscle activity, the lab standard for staging sleep3. All but two studies were crossovers, so people slept once with alcohol and once without.
Sleep has stages. The ones that matter here:
| Term | What it means |
|---|---|
| REM sleep | Rapid eye movement sleep, when most dreaming happens; about 21–30% of a healthy adult's night |
| Deep sleep (N3) | The deepest non-REM stage, also called slow-wave sleep |
| Sleep onset latency | Minutes from lights out to falling asleep |
| REM onset latency | Minutes from falling asleep to the first REM period |
| Wake after sleep onset | Minutes awake after first falling asleep |
Doses were measured in grams of alcohol per kilogram of body mass. The review translated them for an average adult of 62 kg, counting a standard drink as 10 g of alcohol3. Standard drinks differ between countries, so the g/kg figure is the one to trust. Nearly every study gave the alcohol within about 3 hours of bed3.
In plain terms: one night with a measured drink, one night without, both in a sleep lab.
What the studies found
REM sleep fell and came later. On alcohol nights, REM sleep was 11.3 minutes shorter (95% CI −20.8 to −1.8)3. A confidence interval (CI) is the range the true value most likely sits in. The first REM period arrived 18.0 minutes later (95% CI 8.8 to 27.1), and REM's share of the night fell by 2.8 percentage points3.
The more alcohol, the more REM lost. For each extra gram per kilogram, REM fell by a further 40.4 minutes3. The reduction was already significant at a low dose of about 0.50 g/kg, which the authors equate to approximately two standard drinks3.
Falling asleep faster needed a high dose. Overall, time to fall asleep did not clearly change (−2.0 minutes, 95% CI −4.3 to 0.2)3. It shortened only at about 0.85 g/kg, roughly five standard drinks on the paper's scale, by 6.4 minutes per extra g/kg3. An older 2013 review of the sleep literature read it differently, reporting faster sleep onset at all doses1; the 2025 pooled analysis, which could weigh the studies by dose, found it clearly only at high doses. Deep sleep arrived about 3 minutes sooner overall, the dose analysis placed that effect at high doses, and the share of the night in deep sleep did not change3.
The rest was uncertain. Total sleep time was 10.1 minutes shorter on average, but the confidence interval ran from −24.6 to +4.4 minutes3. Time awake during the night and sleep efficiency also did not clearly change. The authors call these results considerably uncertain3.
Alcohol night minus control night, pooled across sleep-lab studies · min
In plain terms: the clearest change was less and later REM, starting at modest doses.
The shape of the night
Alcohol acts in two phases. An older review of the lab literature described the pattern: a more consolidated first half of the night and more disrupted sleep in the second half1. Most studies it reviewed found more slow-wave (deep) sleep early in the night, and a delayed first REM period at all doses1.
The 2025 review gives a likely reason. Peak blood alcohol tends to coincide with sleep onset, and the largest disruption typically comes in the second half of the night, once the alcohol has been broken down3. That second half is normally when most REM happens.
In plain terms: the early night can feel deeper; the later night is where sleep gets lighter and REM is lost.
Your heart overnight
Sleep stages are not the only thing that changes. A 2018 study followed 4,098 Finnish employees in everyday life, comparing each person's nights with and without alcohol2. Heart rate variability (the beat-to-beat variation that rises when the body is relaxed) pointed to less relaxation during the first 3 hours of sleep after drinking2. The study's recovery index fell on average by 9.3, 24.0 and 39.2 percentage units after low, moderate and high intake2. That pattern was similar in men and women, and in active and inactive people2.
Drop in the overnight recovery index, by alcohol intake · percentage units
That study was observational and relied on self-reported drinking, so it shows association within each person rather than a controlled cause.
The nuance
Changed is not the same as ruined. The authors note that REM is normally about 21–30% of a healthy adult's night, so the changes they found "may not elicit clinically meaningful change"3. Whether 11 fewer minutes of REM matters for how you feel the next day was not the question these trials answered.
The sleep-aid idea. The review concludes that any benefit of alcohol for falling asleep "is likely outweighed by greater sleep disruptions to REM sleep across the sleep opportunity"3. The faster sleep onset appeared only at doses where REM loss was larger.
Single nights. Almost every trial tested one night. Regular drinking, tolerance and next-day effects were outside these studies.
Training recovery was not measured. None of the studies here measured strength, muscle or next-day performance. For how poor sleep relates to muscle, see the companion articles on sleep and training.
For any individual question about alcohol or sleep, a doctor is the right person to ask. This page describes studies; it is not advice to drink or not to drink.
In plain terms: alcohol reliably reshapes the night, but how much that matters for any one person is not settled.
Tracking it
Alcohol also counts toward energy intake; see calories in alcohol and why macros don't add up to calories. If training feels flat after a run of short or broken nights, the signs of accumulated training fatigue are worth a look.
The studies at a glance
| Study | Who was studied | Key result | Year |
|---|---|---|---|
| Ebrahim et al., *Alcohol Clin Exp Res*1 | Review of lab studies in healthy volunteers (abstract) | More consolidated first half of the night, more disruption in the second; first REM delayed at all doses | 2013 |
| Pietilä et al., *JMIR Ment Health*2 | 4,098 Finnish employees, nights with vs without alcohol (abstract) | Recovery index during sleep down 9.3, 24.0 and 39.2 units at low, moderate and high intake | 2018 |
| Gardiner et al., *Sleep Med Rev*3 | Meta-analysis of 27 sleep-lab studies, healthy adults 18–70 | REM −11.3 min and delayed 18 min; faster sleep onset only at high doses; total sleep uncertain | 2025 |
| Strüven et al., *Nutrients*4 | 40 healthy adults, smartwatch, 3 nights without, 3 with, 3 after | Resting heart rate during sleep 63.6 → 66.6 → 64.9 beats a minute | 2025 |
Limits worth knowing
- Risk of bias. 21 of the 27 lab studies showed some risk of bias3.
- Small numbers per outcome. The REM duration estimate rests on 96 people3.
- Thresholds are statistical. The "about two" and "about five standard drinks" points mark where effects became significant, not where they start to matter.
- Timing and sex untested. Nearly all alcohol was given within about 3 hours of bed, and there were too few sex-specific data to compare men and women3.
- Watch, not lab. The 2025 heart-rate study used a consumer smartwatch that does not record REM, was not blinded, and had 40 people4.
- Abstract only for Ebrahim 2013 and Pietilä 2018. Gardiner 2025 and Strüven 2025 were read in full.
Latest research
The newest study we opened, published in April 2025, had 40 healthy adults (25 women, 15 men, average age about 30) wear a smartwatch for 9 days4. They had three nights without alcohol, three evenings with the study's dose (40 g a day for women, 60 g for men, as beer or wine), then three nights without. Average resting heart rate during sleep rose from 63.6 to 66.6 beats a minute on drinking nights and fell back to 64.9 after4. The sleep stages the watch could detect did not change significantly. On the questionnaires, 18 of 40 people rated their sleep worse after the alcohol nights4. People knew when they had been drinking, so expectation may have shaped those ratings.
Average resting heart rate during sleep, by phase · beats/min
The short version
- In 27 lab studies, alcohol cut REM sleep by about 11 minutes and delayed it by about 18.
- REM loss appeared at low doses and grew with the amount.
- Falling asleep faster showed up only at high doses; total sleep time was uncertain.
- Overnight heart rate ran about 3 beats a minute higher on drinking nights in a watch study.
Educational overview only — not medical advice.
