Timing Matters More Than the Cup: What Science Says About Coffee and Sleep
It is not how much coffee you drink. It is when you drink the last one.
Everyone has heard someone say "coffee doesn't keep me up, I sleep just fine." It may well be true that they fall asleep — the question is what happens afterwards. And that is where the recent research on caffeine and sleep gets interesting: the effect that matters most is not whether you drift off, but the quality of the deep sleep that follows.
How caffeine works, in plain terms
Caffeine blocks adenosine receptors — adenosine being the compound that accumulates in the brain through the day and signals "time to rest." Without that signal, the body takes longer to register fatigue. The effect does not fade quickly: caffeine's half-life is roughly 5 to 6 hours, which means a good share of what you drank at 4pm is still circulating at 10pm.
What the studies actually show
A systematic review and meta-analysis published in Sleep Medicine Reviews (Gardiner and colleagues, 2023), pooling 24 controlled studies, found that caffeine cuts total sleep time by around 45 minutes, reduces sleep efficiency by 7%, and adds 9 minutes to the time it takes to fall asleep. More striking: it reduces the proportion of deep sleep (N3, the stage tied to physical recovery) and increases light sleep (N1). From that data, the authors calculated that a cup of coffee (107 mg of caffeine) would need to be consumed about 8.8 hours before bed so as not to shorten sleep.
That is the number that circulates online. But the same research group went further — and the result complicates the rule.
In 2025, in the journal Sleep, Gardiner and colleagues published a randomised clinical trial, crossover and double-blind, testing two doses (100 mg and 400 mg) at three timepoints: 12, 8 and 4 hours before bed. The finding that matters to anyone who enjoys an afternoon cup: at 100 mg — the equivalent of one cup — there was no significant effect on sleep at any of the three timepoints, not even four hours before bed. The disruption showed up at 400 mg, which cut deep sleep by 15 to 30 minutes and sleep efficiency by 7% to 10%.
Two caveats, and they pull in opposite directions:
The meta-analysis pools studies with varying doses and protocols, and the 8.8-hour figure is a statistical extrapolation, not a limit measured person by person. The 2025 trial, by contrast, measured precisely — but in 23 healthy men averaging 25 years of age, a small and narrow sample that does not speak for someone who is 60, for women, or for anyone who is genetically slow to metabolise caffeine. Neither one hands you a personal rule.
What this changes in practice
This is not about cutting out coffee — and, reading the two studies together, what matters is dose and timing together, not timing on its own. A cup after lunch has direct experimental support for being harmless to most people's sleep. What deserves attention is accumulation: three or four cups concentrated from the afternoon onward easily add up to the 400 mg the trial showed to be disruptive — and at that point timing genuinely starts to weigh.
And there is a way out for anyone unwilling to give up the afternoon cup: the ritual does not depend on the caffeine. A coffee with a well-defined origin, brewed slowly in a method like the French press, delivers aroma, body and the same pause at the end of the day — and the decaffeinated version of a specialty bean preserves much of that. What moves is the caffeine, not the habit.
It is worth remembering that cup quality also affects how much of it you need: a poorly extracted coffee, bitter through brewing error, tends to invite a second serving that might not have been necessary.
In the end, understanding your own caffeine curve is less about discipline than about observation: noticing how you sleep on the days the last cup came earlier, compared with the days it did not. That is practical information anyone can gather alone — long before any study, every cup was already a choice.