Caffeine and sleep: the half-life math no one teaches you.

The most common mistake adults make about caffeine is also the most quietly costly: confusing the ability to fall asleep after an afternoon coffee with the absence of an effect on sleep. The two things are not the same. Falling asleep is a behavioural milestone. What happens during the next eight hours — the architecture of the sleep, the cycling between stages, the amount of slow-wave and REM you actually get — is the part that determines whether tomorrow's brain functions properly. And on that, caffeine has a great deal more to say than most habitual drinkers realise.

§ The mechanism: caffeine does not give energy

It is worth starting with the pharmacology, because the popular framing is wrong in a way that matters. Caffeine does not provide energy and does not stimulate the brain. It blocks a signal of tiredness.

Throughout the day, every cell in the brain accumulates adenosine as a byproduct of energy use. Adenosine binds to A1 and A2A receptors, and as the binding increases over the course of the waking day, you feel more tired. This is the primary biochemical signal of sleep pressure. At a sufficient threshold, paired with circadian readiness, it produces sleep.

Caffeine is structurally similar enough to adenosine that it occupies the same receptors without activating them. The adenosine is still being produced — it is being made all day — but the brain cannot read it. You feel alert. The caffeine is not adding wakefulness; it is hiding the accumulated tiredness.

This matters for sleep because, when the caffeine clears, the adenosine is suddenly readable again at the concentration it has built to. This is part of why a 3 p.m. coffee produces both an afternoon of alertness and a curiously sudden crash in the evening: the caffeine clears just enough for the underlying signal to be partially read again.

§ Half-life: the number that decides everything

Caffeine's average half-life in adults is approximately five hours. That figure obscures a wide range: the typical adult sits between three and nine hours, and the variation is largely genetic, driven by polymorphisms in the CYP1A2 enzyme responsible for caffeine metabolism. Pregnancy, oral contraceptives, certain medications and liver function further modulate clearance.

Five hours, applied honestly, means a 200 mg coffee (roughly a strong drip cup) consumed at 3 p.m. leaves 100 mg in your system at 8 p.m. and 50 mg at 1 a.m. For a slow metaboliser the picture is significantly worse — 50 mg might still be in circulation by 3 or 4 a.m.

Fifty milligrams is not nothing. It is, roughly, the dose of a strong cup of black tea. People do not drink that at midnight by choice. They consume it inadvertently because the math of the half-life is not intuitive.

"The question is not whether you can fall asleep after the afternoon coffee. The question is what the residual caffeine, still occupying your adenosine receptors at 2 a.m., is doing to the architecture of the sleep you are having."

§ What the controlled sleep studies show

The most cited piece of work here is Christopher Drake's 2013 study, in which 400 mg of caffeine (roughly two cups of brewed coffee) was given to subjects at 0, 3, and 6 hours before scheduled bedtime, with sleep measured against placebo using both subjective reports and objective home polysomnography.

All three conditions degraded sleep relative to placebo. Even the six-hour-before-bed dose — that 5 p.m. coffee for a 11 p.m. bedtime — reduced total sleep time by more than an hour and reduced sleep efficiency measurably. Subjects' subjective reports often missed the effect. They felt they had slept fine. The recording said otherwise.

A broader systematic review by Clark and Landolt in 2017 confirmed the pattern across more than a hundred studies: caffeine consumed within six hours of bedtime reduces total sleep duration, increases sleep latency (time to fall asleep), and reduces deep slow-wave sleep — which is the stage most associated with overnight memory consolidation, hormonal regulation, and the subjective feeling of having actually rested.

The dose-response curve is gentle but unambiguous. More caffeine, closer to bedtime, equals worse sleep. There is no robust evidence of tolerance to the sleep-disrupting effects, even in habitual users — tolerance to the subjective alerting effect is real, but tolerance to the architectural disruption of sleep appears to be limited or absent. Heavy daily drinkers stop feeling the caffeine. They do not stop having their sleep degraded by it.

§ The slow-metaboliser problem

Approximately half the adult population carries at least one copy of a CYP1A2 variant that slows caffeine clearance. Genetic testing is the only way to know definitively, but there are practical signals: if afternoon coffee gives you jitters disproportionate to the dose, if you find that even a morning coffee occasionally affects sleep onset, if you have ever noticed that you tolerate caffeine markedly worse than a friend at the same dose — you are likely a slow metaboliser.

For slow metabolisers the half-life can be seven to nine hours. A 10 a.m. 200 mg coffee can still have 50 mg circulating at midnight. The 'no caffeine after noon' rule is, for these people, a conservative floor and not a hard ceiling.

§ The practical protocol

Bringing the pharmacology back to daily decisions, three rules cover the great majority of cases:

• →Set a personal cutoff and respect it. For most adults, noon is conservative; 2 p.m. is the latest defensible time for a 10 or 11 p.m. bedtime. Slow metabolisers should consider 10 a.m.
• →Treat the dose as fixed. Two cups of coffee before noon is a normal protocol. A third cup at 4 p.m. is the one that does the damage, because the math is unforgiving.
• →Notice the architecture, not the onset. If you fall asleep fine but wake at 3 a.m., if you sleep eight hours and wake unrested, if your subjective sleep quality has drifted downward over months — caffeine timing is among the first things to audit, ahead of more exotic explanations.

§ A closing point on honesty

Most adults who drink caffeine know, somewhere, that the afternoon coffee is doing them no favours. The behaviour persists because the cost is invisible — it is paid in a slightly worse sleep that you have nothing to compare against, in a slightly worse next-day cognition that you attribute to other things, in a slow upward creep in evening caffeine to compensate for fatigue that the same evening caffeine is producing.

The pharmacology is not contested. The half-life is what it is. The simplest single change you can make to the quality of your sleep, with the lowest behavioural cost, is to push the day's last caffeine earlier by two hours. The bank does not negotiate, and the receptors do not lie.