When to stop drinking coffee: the half-life nobody accounts for

A mid-afternoon coffee is still working at bedtime. Caffeine costs about 45 minutes of sleep on the nights people drink it, and the safe gap for a regular coffee works out at 8.8 hours before bed.¹ Go to bed at eleven and that is a two o'clock cut-off. Early afternoon, not evening.

The half-life is not a number you can plan around

Advice about coffee almost always rests on one figure: the half-life, the time the body takes to clear half of a dose. It gets quoted as though it were a fixed property of the molecule. It is a property of the person, and it shifts with how much that person drank.

Twelve volunteers took a placebo, then 250 mg of caffeine, then 500 mg, on separate occasions. At the larger dose the body cleared caffeine more slowly, and the half-life stretched out.² Doubling the dose did more than double how long it stayed.

In cups, that is the difference between a couple of coffees and about five. A 250 mL coffee carries roughly 107 mg of caffeine, so a 500 mg day is a five-cup day.¹ The later cups do not simply add to the earlier ones. They slow the exit of everything already circulating.

So a rule of the form "stop drinking six hours before bed", derived from one half-life value, is aimed at a moving target. The target moves the wrong way on exactly the days you drank the most.

What caffeine does to a night

Set caffeine against a placebo and measure the night that follows. Total sleep time drops by 45 minutes.¹ Sleep takes 9 minutes longer to arrive, and people spend 12 more minutes awake in the middle of the night.¹

The shape of the night changes as well as its length. Deep sleep falls by 11.4 minutes, and lighter sleep takes its place.¹ That is why a night after a late coffee can look long enough on paper and still feel thin in the morning.

The timing trials are the useful part

One trial went at the clock directly. It gave people 400 mg of caffeine, around four cups of drip coffee, either at bedtime, 3 hours before bed, or 6 hours before bed.³ Sleep was recorded at home with a monitor rather than by asking people how they had slept. All three timings disturbed sleep, and the six-hour gap was still too short.³

The trial data let you run that arithmetic backwards, from a dose to an hour on the clock. A 250 mL coffee holding 107 mg of caffeine needs 8.8 hours before bed. A standard scoop of pre-workout supplement, at 217.5 mg, needs 13.2 hours.¹

Hours before bed to stop, by sourcehours before bedtime

Standard coffee (107 mg)8.8 h

Pre-workout serving (217.5 mg)13.2 h

Put those on a real day. Bed at eleven means the last coffee lands around two in the afternoon, and the last pre-workout around ten in the morning. An evening gym session with a scoop of pre-workout is a decision about the night, not only about the training.

Why two people can drink the same coffee and disagree

Caffeine is broken down mainly by one liver enzyme, CYP1A2, and a common variation in the gene for it changes what a given dose does. Competitive cyclists were given 4 mg of caffeine per kilogram of body mass, which is about 280 mg for someone who weighs 70 kg, or a little under three cups. On average they rode 3% faster than on placebo.⁴ Riders carrying one particular version of the gene rode 13.7% slower on the identical dose.⁴

That is a cycling trial, not a sleep trial, and no bedtime can be read out of it. What it establishes is narrower and still worth having: the same dose is not the same exposure in two different people. The friend who drinks espresso after dinner and sleeps fine may not be more hardened to caffeine. He may simply be clearing it faster.

What people report about themselves points the same way, with a twist. Among young adults, caffeine use went with more disturbed sleep, but with no measurable change in how long they slept.⁵ The caffeine users also tended to be night owls. Cause and effect run in both directions here. A late body clock invites a late coffee, and the late coffee pushes the clock later still.

The dose hiding in things that are not coffee

Coffee is the drink people count, and it is often not the largest dose of the day. A scoop of pre-workout carries about twice the caffeine of a cup, which is why its cut-off falls in the morning rather than the afternoon.¹

Energy drinks are studied less carefully as a timed dose and linked more clearly to harm. People who drank them had roughly five times the odds of insomnia, and more than three times the odds of jitteriness, compared with people who did not.⁶ Those are odds from observation, not the effect of a known dose at a known hour, so no timing rule falls out of them. They do suggest the format matters. A can goes down fast, and usually late.

What remains unknown

The cut-offs are population averages built on average clearance. Nobody has run a trial that sets a bedtime caffeine limit by genotype, so 8.8 hours is the right default and the wrong personal answer for some unknown share of readers.¹

The sleep estimates come mostly from short studies in people who were handed caffeine, not from habitual drinkers followed for months. Whether regular use dulls the effect on sleep the way it dulls the jitters is not settled here. And the dose-dependence of clearance was measured at 250 and 500 mg, a range that brackets a heavy day but leaves the ordinary two-cup dose untested.²

The honest summary is narrower than the usual advice and more useful for being narrow. The half-life is not one number, so stop planning around it as though it were. Use the clock instead: coffee by early afternoon, pre-workout by mid-morning, and a late can of anything treated as what it is.