Why alcohol puts you to sleep and then wakes you at 3 a.m.

A drink before bed can shorten how long it takes to fall asleep, but only at a fairly high dose. That same dose damages the sleep that follows.¹ The trade gets worse as the dose rises. The drink that helps you drop off is also the one most likely to have you awake in the small hours.

What "helps you sleep" actually means in the lab

Sleep researchers measure two separate things. One is how fast someone falls asleep. The other is what the rest of the night looks like once they do. A review that pooled 27 studies on healthy adults found these two outcomes respond to different amounts of alcohol.¹ The review found alcohol delayed the start of REM sleep and cut its length. A low dose, roughly two standard drinks, was already enough to disrupt REM sleep, the stage tied to dreaming and memory processing.¹ Falling asleep faster and reaching deep sleep sooner only showed up at a high dose, roughly five standard drinks. By then, REM disruption had gotten worse too.¹

That is the shape of the trade. There is no dose in this review that speeds up sleep onset without a REM cost attached. The two effects arrive together, and the second one only gets stronger as the first one kicks in.

The first half of the night is not the whole story

A polysomnography study gave healthy volunteers alcohol at bedtime and recorded their brain activity through the night. At a blood alcohol level of 0.1%, alcohol behaved like a short-acting sleep drug for the first several hours. People fell asleep faster. They woke up less. They spent less time in the lightest stage of sleep, and more in slow-wave sleep, the deepest stage, early in the night.² But in the same night, during the second half, the study already picked up rebound effects, including more of that light stage returning.² The lower dose tested, a blood alcohol level of 0.03%, showed no clear effects, though the study was small.²

That rebound is a plausible explanation for a rough second half of the night after a nightcap, though the study did not track waking at 3 a.m. itself. The body clears alcohol over the course of the night. Once blood alcohol falls, sleep does not simply continue as it would have without the drink. It shifts toward lighter stages, in the same session where it had looked improved a few hours earlier. The apparent benefit and the later cost happen back to back, inside one sleep period, not on separate nights.

Two more links between alcohol and worse sleep

Both sit apart from the fall-asleep effect.

The first is breathing. A meta-analysis of studies using overnight breathing monitors found that alcohol raised the apnea-hypopnea index, a count of breathing pauses per hour.⁴ The pooled difference was nearly 4 more events an hour compared with no alcohol.⁴ The lowest oxygen level reached during the night also ran about 2.7 percentage points lower.⁴ More pauses in breathing mean more brief arousals, whether or not the sleeper remembers waking the next morning.

The second is heavier, long-term drinking. In people with alcohol use disorder, wrist-worn sleep trackers recorded total sleep time running about 45 minutes shorter on average than in people without the disorder.³ The pooled difference was -44.67 minutes, and the interval around it stayed well below zero.

Total sleep time in alcohol use disorder, wrist-tracker studies0 = no effect

Total sleep time, in minutes-44.67 (-56.10 to -33.24)

That is a different population from someone having an occasional evening drink. It shows what heavier, regular drinking is associated with, not what one dose does to one night.

Dose and timing change the trade, but only partly

Put the numbers from the dose-response review next to each other and a rough map appears. Around two drinks, REM sleep starts taking a hit, with no benefit yet to falling asleep faster.¹ Around five drinks, falling asleep and reaching deep sleep both speed up, but REM disruption is now worse than it was at the lower dose.¹

Timing matters by the same logic, even though none of these studies set out to test a specific number of hours before bed. The rebound pattern documented at bedtime doses of alcohol shows up within the same night, once blood alcohol starts falling.² It is reasonable to guess that an earlier drink would move that window, but no trial here tested it.

What remains unknown

None of the studies behind this article measured a specific number of hours to stop drinking before bed. Any such number is a guess, not a finding. The dose-response review also flagged large uncertainty in its own totals for overall sleep time, sleep efficiency, and total time spent awake after falling asleep, which is exactly the kind of whole-night summary a person actually cares about.¹

The lab studies used single, controlled doses in supervised settings. An evening of mixed drinks at an uncontrolled pace is not the same thing, and none of these trials modeled it directly. Most of this evidence also comes from adults without a diagnosed sleep or alcohol problem. What changes for someone who already has one is a separate picture, and in the case of alcohol use disorder, a considerably worse one.³

What holds across the studies is the shape of the trade-off itself. A fast route to sleep onset comes bundled with a rougher second half of the night, at every dose tested where the fast route appeared at all.^1,2