Most nights carry two familiar defaults. A glass of something, a plate of something, both close enough to bedtime that they compete for blame when morning arrives flat. Between a late-night drink and a late-night snack, the gut check runs fast: both probably cost you sleep, both probably matter about the same, both probably deserve the same low-grade guilt.
That scoring system holds together only when the evidence stays vague. When someone wires up the instruments that track what sleep does across a full night, one half of this matchup drops out entirely.
Is a Late-Night Drink or a Late-Night Snack Worse for Your Sleep?
The late-night drink is worse for your sleep, and it is not close. Two standard drinks delay REM sleep onset by 18 minutes. Late eating, measured with polysomnography in a controlled crossover trial, produced zero difference in any sleep parameter. The snack does have a cost, but it targets hunger and metabolism the next day, not tonight's sleep.
— Gardiner et al. 2024 · Sleep Medicine Reviews · 27 studies, n=529 | Vujovic et al. 2022 · Cell Metabolism · n=16
When researchers gave the same people early meals and late meals on separate weeks, with sleep monitors tracking brain waves through every stage, the late eating scored nothing. Same total sleep. Same time in each stage. Same architecture in every measurable parameter. Moving identical meals four hours later changed hunger, changed metabolism, changed fat-storage gene activity. Sleep registered zero difference.
The drink was not so clean. Two standard drinks delayed the onset of REM sleep by 18 minutes. Not five drinks. Not a binge. Two glasses of wine, the pour most people default to at dinner. REM is where the brain consolidates what it learned that day, and its disruption begins at a dose that barely qualifies as a night out.
BLAMED: The late-night snack — digestive effort that registered zero on every sleep metric
ACTUAL: The late-night drink — REM onset delayed 18 minutes at a two-glass pour
The reason most people never notice: the opening hours of sleep genuinely are deeper after alcohol. Heavier slow-wave sleep, the kind that feels like falling into a wall. The brain does sink further during those early hours, and the morning check reads fine. The disruption hides in the second half, in stages your morning self-assessment has no way to register.
The damage extends past sleep. When REM gets compressed, the brain's appetite controls shift the next day. The parts that normally brake food desire go quiet. The parts that amplify wanting get louder. The net result: roughly 385 extra calories consumed the following day, aimed at the most calorie-dense options, without any increase in actual hunger. The drink costs you sleep tonight and self-control tomorrow.
The threshold matters. At one drink, the measurable disruption has not started. At two, it begins. This gap runs between two drinks at dinner and a plate of food before bed, not between one glass and a snack. The dose draws the line.
The snack walked away clean on sleep. On appetite, the picture flips. The same trial that found zero sleep difference found that eating later made next-morning hunger twice as likely. Identical meals, identical portions, different clock, different appetite. The late-night snack does not damage your sleep. What it does to the appetite machinery running the next day belongs to a separate ledger.