Two drinks is enough to disrupt REM sleep; five just makes it worse

Alcohol is one of the most common things people reach for before bed, and it is often consumed in the belief that it helps sleep. That belief has a grain of plausibility, since alcohol is sedating, but sedation and healthy sleep are not the same thing. What has been less clear is exactly which parts of the night alcohol changes, and whether the amount you drink matters. This review set out to answer both questions, looking specifically at how the dose and timing of alcohol intake influence the characteristics of night-time sleep in healthy adults.

The researchers ran a systematic search of the published literature and identified 27 studies suitable for pooling. All examined the effect of alcohol on subsequent night-time sleep in healthy adults. The analysis looked at sleep architecture, meaning the structure of the night: when rapid eye movement, or REM, sleep began and how long it lasted, how quickly people fell asleep, how quickly they reached deep non-REM stage three sleep, plus total sleep time, sleep efficiency and time spent awake after first falling asleep. Doses were standardised by body weight so that low and high intakes could be compared across studies.

Alcohol changed the structure of sleep: the onset of REM sleep was delayed and its total duration reduced. A dose-response relationship emerged, with REM disruption appearing after a low dose of 0.50 grams per kilogram of body weight or less, roughly two standard drinks, and progressively worsening as the dose increased. Falling asleep faster, and reaching deep N3 sleep sooner, was only observed after a high dose of 0.85 grams per kilogram or more, roughly five standard drinks. For total sleep time, sleep efficiency and wake after sleep onset, the pooled results carried too much uncertainty to draw a conclusion either way.

This analysis directly challenges the nightcap belief that inspired it. It confirms earlier suggestions that alcohol reshapes sleep architecture, and it adds precision: the REM cost begins at doses many people would consider modest, while the fall-asleep-faster effect only appears at heavy doses and likely exacerbates the REM disruption. The authors flag clear open questions, including how personal and environmental factors that affect alcohol metabolism change the picture, and whether the effects differ between men and women. Timing of intake relative to bedtime also remains an area where the evidence could be sharpened.

This is a reason to think that a drink or two in the evening changes the quality of your sleep even if it does not obviously change how long you sleep. REM sleep is part of a normal night, and this analysis suggests it is the part alcohol reliably erodes, starting at around two standard drinks. It is also a reason to be sceptical of the feeling that heavy drinking helps you sleep: dropping off faster after five drinks does not mean the night that follows is a good one. How this plays out for any individual will depend on factors the research has not yet pinned down, including sex and how quickly a person metabolises alcohol.