Caffeine Half-Life: Why Your Afternoon Coffee Is Still There at Midnight

One equation, and what it rules out

Caffeine follows first-order elimination at the doses people actually drink. That phrase has a precise meaning: the body removes a constant proportion of what is present per unit time, not a constant number of milligrams. Twice the caffeine on board means twice the caffeine cleared per hour, which is why the curve is an exponential and not a straight line down to zero. The remaining amount is the dose times one half raised to the power of elapsed time divided by half-life, written 200 × (1/2)^(8/5) for a 200 mg dose eight hours ago at a five-hour half-life. That expression is the whole model, and every number in this article comes out of it.

What the equation rules out is just as useful. It has no term for how tired you feel, no term for tolerance, and no floor at which the drug is declared gone — the curve approaches zero and never touches it. It also has no term for anything you might do to speed it up. Water, food, exercise and cold showers do not appear, because none of them meaningfully changes hepatic clearance. The two variables that matter are the dose you took and the half-life your liver runs at, and only one of those is under your control.

Half is still half: the hour-by-hour decay of one cup

Start at 3 p.m. with 200 mg and a five-hour half-life. At 4 p.m. you have 174 mg, at 6 p.m. 132 mg, at 8 p.m. exactly 100 mg — one half-life, half the dose. By 11 p.m., eight hours in, 66 mg is still circulating, which is about the caffeine of a shot of espresso taken at bedtime. You cross below 50 mg at 1 a.m., below 25 mg at 6 a.m., and are still carrying 12 mg when you wake up. The FDA puts an 8-ounce cup of drip coffee at roughly 80 to 100 mg, so what remains at midnight from a single afternoon coffee is not a trace: it is a meaningful fraction of another cup.

The trap in the phrase "half-life" is that people hear it as a lifetime. One half-life removes 50%, two remove 75%, three 87.5%, four 93.75%, five 96.9%. Clinicians conventionally treat four to five half-lives as the point at which a drug is effectively gone, so a five-hour half-life means twenty to twenty-five hours, not five. That is why a cup at 3 p.m. is still measurable at breakfast, and why someone who drinks coffee from waking until mid-afternoon never actually reaches zero on a working week.

The half-life is a range, and the range has causes

The Institute of Medicine's review of caffeine pharmacology gives a mean plasma half-life of about 5 hours in healthy individuals and a range of roughly 1.5 to 9.5 hours. That is more than a sixfold spread, and it is the single most important fact on this page — larger than any difference between a strong and a weak coffee. Most of it traces to one enzyme. CYP1A2 handles the great majority of ingested caffeine, and its activity varies between people both genetically, through common polymorphisms in the CYP1A2 gene, and environmentally.

Three environmental shifts are large enough to change the answer rather than nudge it. Smoking induces CYP1A2, and the clinical literature summarised in StatPearls puts the reduction in half-life at up to 50% in smokers — a smoker's afternoon coffee really is a different drug from a non-smoker's. Combined oral contraceptives run the other way: the Institute of Medicine review states plainly that their use can double caffeine half-life. Pregnancy is the largest single effect, with the final trimester associated with half-lives of up to 15 hours as CYP1A2 activity falls. None of these are exotic. Between a smoker and a woman in her third trimester, the same espresso persists for radically different lengths of time.

Two people, one cup, 3.6 times the drug at bedtime

Put two people at the same table at 3 p.m. and give both 200 mg. One smokes; call their half-life 3 hours. The other takes a combined oral contraceptive; call theirs 10 hours, which is inside the doubling the Institute of Medicine describes. At 11 p.m. the first has 31 mg on board and the second has 115 mg — a factor of 3.6 from a single identical cup. Neither of them has done anything wrong, neither has a stronger coffee, and neither can tell from the taste. This is what people are actually reporting when they say caffeine "does not affect" them or that they are "very sensitive" to it: often it is not receptor sensitivity at all, it is clearance.

The practical consequence is that a cut-off hour cannot be copied from someone else. "No coffee after 2 p.m." is a rule derived from a five-hour half-life and an eleven o'clock bedtime, and it is simply wrong for the fast and the slow ends of the distribution. The way to find yours is not a genetic test but a subtraction: pick a residue you are willing to go to bed with, then work backwards from your bedtime using the half-life the calculator lets you set, and adjust it over a fortnight against how you actually slept.

Doses stack: the afternoon cup lands on the morning ones

Exponential decay is additive, so a day of coffee is just several curves summed. Take a common pattern: 100 mg at 8 a.m., 100 mg at 10 a.m., 200 mg at 1 p.m. and 200 mg at 4 p.m., 600 mg in all — above the 400 mg a day the FDA cites as not generally associated with negative effects, but not unusual. At a five-hour half-life, what is still circulating at 11 p.m. is 13 + 16 + 50 + 76 = 155 mg. At a three-hour half-life it is 68 mg; at nine and a half hours, 289 mg. The bedtime residue moves by more than a factor of four across the same day's drinking, purely on clearance.

Notice which term dominates. The 4 p.m. dose contributes 76 mg of the 155 mg, the 1 p.m. dose 50 mg, and the two morning cups together only 29 mg. That is not because the morning coffees were smaller — it is because they have had fifteen and thirteen hours to decay while the afternoon ones have had seven and six. Cutting the last cup of the day is worth several times more, at bedtime, than cutting the first, which is the single most actionable thing the arithmetic tells you.

Feeling awake is not the same as having cleared it

The subjective read on caffeine is unreliable in a specific, measurable way. Drake and colleagues gave healthy sleepers a fixed 400 mg dose at 0, 3 or 6 hours before their habitual bedtime and measured sleep both by self-report and by a validated home monitor. Even at six hours before bed, objectively measured total sleep time fell by more than an hour — and the participants largely did not notice. Their subjective reports failed to register the disruption the device recorded. That gap is the reason a personal rule built on "I sleep fine after an evening coffee" is built on the wrong evidence.

The reason the two diverge is that alertness and blood concentration are different curves. Tolerance blunts the felt stimulation over weeks of daily use without changing hepatic clearance at all, so a habitual drinker can genuinely stop feeling a cup while still carrying the same milligrams at the same hour. The calculator tracks the milligrams. Your body clock does not care whether you enjoyed them.

Adenosine: the debt keeps accruing while the alarm is muted

Caffeine does not supply energy. It crosses the blood-brain barrier and acts as an antagonist at adenosine receptors, blocking a signalling molecule that accumulates through the waking day and whose build-up is one of the physiological correlates of sleep pressure. Blocking the receptor does not remove the adenosine. It stops you reading the gauge. That is the mechanistic version of a very ordinary experience: the tiredness you were not feeling at 6 p.m. arrives all at once when the block wears off.

This is why the decay curve matters more than how you feel. As the concentration falls through the night, receptors free up while the accumulated adenosine is still there, and the pressure that was masked at 11 p.m. asserts itself in the small hours — which is one mechanism by which caffeine fragments the second half of the night rather than merely delaying its start. Calculating what remains at bedtime is not fussiness. It is the only part of this system you can see before you get into bed.

Frequently asked questions

If the half-life is 5 hours, is the caffeine gone after 5 hours?

No — half of it is gone. From 200 mg you still have 100 mg. Two half-lives leave 25%, three leave 12.5%, four leave 6.25%. Clinicians conventionally treat four to five half-lives as effective clearance, so a five-hour half-life means twenty to twenty-five hours before the dose is negligible, not five.

What time should I stop drinking coffee?

There is no universal hour, because the half-life is not universal. Do it backwards instead: decide what residue you are willing to go to bed with, then use the calculator with a half-life you have reason to believe applies to you. At a five-hour half-life and an 11 p.m. bedtime, a 200 mg coffee leaves 66 mg — cutting it to 100 mg or moving it two hours earlier both roughly halve that.

I fall asleep fine after an evening coffee. Does that mean it does not affect me?

Not on the evidence. In Drake and colleagues' 2013 trial, 400 mg taken six hours before bed cut objectively measured total sleep time by more than an hour while participants' own reports largely missed the disruption. Falling asleep and sleeping well are different outcomes, and caffeine damages the second more reliably than the first.

Does drinking coffee every day make me clear it faster?

Habitual use mainly changes what you feel, not how fast you clear. Tolerance blunts the subjective stimulation over weeks; hepatic clearance is set largely by CYP1A2 activity, which responds to things like smoking, hormonal contraception and pregnancy rather than to how much coffee you drink. So a heavy daily drinker can stop noticing a cup while carrying exactly the same milligrams at bedtime.

Why does the same coffee affect my partner so much more than me?

Most likely because you clear it at different rates. Published adult half-lives run from about 1.5 to 9.5 hours, and CYP1A2 — the enzyme that handles the great majority of ingested caffeine — varies genetically and is induced by smoking and slowed by combined oral contraceptives. Eight hours after an identical 200 mg cup, a 3-hour clearer has 31 mg left and a 10-hour clearer 115 mg.

I am pregnant, or taking the combined pill. What does the longer half-life mean in practice?

It means the same intake produces a higher standing level, because each dose has less time to clear before the next one. EFSA's 2015 opinion recommends that pregnant and lactating women limit caffeine to 200 mg a day from all sources, against 400 mg for other adults; late pregnancy has been associated with half-lives of up to 15 hours. This is a conversation for your midwife or doctor, not a web page.