Caffeine has a half-life, and that is the whole problem

This looks at caffeine and sleep from the practical end — what holds up once conditions stop being ideal.

Half-life means it lingers

Caffeine has a half-life commonly cited at around five hours, so half of an afternoon dose is still circulating at bedtime. That range is wide between individuals — genetics, liver enzyme activity, pregnancy and some medications all shift it substantially.

Someone metabolising slowly can have a meaningful proportion of a lunchtime coffee present at midnight. Smoking speeds the clearance considerably, which is why people who give up frequently find that their usual intake has suddenly become too much.

Falling asleep is not the measure

Many people who fall asleep easily after evening caffeine still show reduced deep sleep on measurement. That produces a night that feels adequate and is less restorative, which is difficult to notice from the inside. This is why "caffeine does not affect me" is a claim about sleep onset rather than about sleep quality.

Tolerance also builds to the alerting effect faster than to the sleep disruption, so a habitual drinker ends up getting less lift for the same cost.

The mechanism

Caffeine blocks adenosine receptors, and adenosine accumulation through the day is what produces sleep pressure. Blocking it does not remove the accumulated adenosine; it masks it, which is why the tiredness arrives abruptly when the caffeine clears. That rebound is the reason the afternoon coffee often leads to another.

For most people, regular use appears to increase the number of adenosine receptors, which is what turns the morning coffee from a lift into the thing that returns you to baseline.

A cut-off is the practical tool

Setting a personal cut-off — commonly early afternoon — is simpler than calculating doses. Moving it earlier for two weeks and noticing whether sleep changes is a straightforward personal experiment. For people with disrupted sleep, this is among the highest-yield changes available.

Two weeks is the minimum honest test, because the first few days of an earlier cut-off include withdrawal and will make the change look worse than it is.

Withdrawal is real and short

Reducing intake produces headache, fatigue and irritability for a few days in regular users. Tapering over a week rather than stopping abruptly makes it manageable.

None of this is an argument against caffeine, which has real benefits — it is an argument about timing. Pregnancy, some heart rhythm conditions and a number of medicines are the situations where the amount matters as much as the hour, and those are questions for a clinician or pharmacist rather than for a general rule.

Adjust the size of it until it is something you would actually do tired.

The dose is less predictable than the drink suggests

Caffeine content varies widely with bean, roast, grind and brewing method, so two drinks described the same way can differ several times over. A single espresso contains less than a large filter coffee despite tasting far stronger, because the serving is tiny. Tea, cola, energy drinks, chocolate, some supplements and a number of over-the-counter painkillers all contribute, and the painkillers are the ones people miss entirely.

Anyone working out a cut-off is therefore better off counting sources than counting cups.

The takeaway

Set a cut-off time rather than counting cups. Half of it is still there five hours later.

Pick the one that costs you least, and let the rest wait.