Your Caffeine Cut-Off Isn't a Time, It's a Half-Life
The usual advice is a clock time: no coffee after two, or after three, pick your rule. It is the wrong shape of advice, because caffeine does not care what time it is. It clears on a half-life, and a half-life means the answer depends on the dose you had and on how fast your particular body processes it.
What "half-life" actually means here
Caffeine leaves the body by first-order decay: a fixed proportion disappears per unit time, not a fixed amount. With a typical adult half-life of about five hours, 200mg becomes 100mg after five hours, 50mg after ten, 25mg after fifteen.
That proportional shape is why the "one cup won't matter" intuition fails. Doubling the dose does not add a fixed penalty — it adds a whole half-life to the time needed to get back down to any given level.
Four ordinary cases
Run through the caffeine timing calculator, using a 50mg threshold as the level unlikely to interfere with sleep:
- A filter coffee (95mg), 8 hours before bed — 31.3mg still circulating at bedtime, a third of the dose. Under the threshold.
- A large coffee (200mg), 6 hours before bed — 87.1mg at bedtime, 43.5% of it. Over. That dose needs 10 hours to fall to 50mg.
- A double espresso (80mg), 10 hours before bed — 20mg left. Comfortably clear.
- An energy drink (160mg) at 4pm for a midnight bed — 91.9mg at bedtime, 57.4% of the dose still there. It needs 8.4 hours, so 4pm was already too late.
Notice what separates these. It is not the time of day at all. It is dose size against hours remaining.
Why a fixed cut-off time fails
A 2pm rule serves a 95mg filter coffee and an 160mg energy drink identically, though one needs 4.6 hours to clear the threshold and the other needs 8.4. If you go to bed at 10pm, the first is fine from 5:30pm onward and the second was borderline at lunchtime.
The rule also ignores when you actually sleep. Somebody who goes to bed at 9pm and somebody who goes at 1am have four hours of difference to work with — a difference worth roughly one whole halving of whatever they drank.
The useful version is not a time but a subtraction: take your bedtime, subtract the hours the calculator gives for your dose, and that is your cut-off today. On a day you have a bigger coffee, the cut-off moves earlier by itself.
The threshold is a convention, not a fact
The 50mg figure is a reasonable default and nothing more. Sensitivity to caffeine at bedtime varies substantially between people, and some notice effects well below it while others sleep through considerably more.
The calculator lets you change it, and the honest way to set it is empirically: if you are sleeping badly, lower it and see whether that helps before concluding caffeine is not your problem. If you sleep well on a hundred milligrams, the default is being harsher than you need.
What this model leaves out
Being clear about the limits, because the arithmetic is tidier than the biology:
- Absorption is not instant. The model treats the dose as arriving all at once; in practice caffeine peaks roughly half an hour to an hour after drinking. That makes the estimate mildly conservative.
- Half-life is not a constant. Five hours is a population average. Yours could be meaningfully shorter or longer, and the difference dominates everything else here.
- Doses stack. A morning coffee is still partly present when the afternoon one arrives. The calculator handles one dose at a time; for a heavy day, work from the last significant one and treat the result as a floor.
- Sleep is not only about caffeine. Clearing the threshold does not guarantee a good night, and being over it does not guarantee a bad one.
This is a planning estimate, not medical advice. Persistent trouble sleeping is worth raising with a clinician rather than solving with arithmetic, and if you are pregnant, taking medication, or managing a heart condition, your caffeine question is a question for them and not for a calculator.
The habit worth building
Work out the cut-off once for the two or three drinks you actually have — your usual morning coffee, the afternoon one, whatever you reach for when tired — and remember those three numbers rather than a rule.
Most people find the answer is earlier than they assumed, and that the surprise is not the morning coffee but the innocuous-looking one at four o'clock. If you want to see how much that costs across a week rather than a night, the sleep debt calculator puts a number on the accumulated shortfall.
How much is actually in your drink
The arithmetic is only as good as the dose you feed it, and this is where most people's estimate goes wrong. Rough figures, and they are rough:
- Filter coffee, a mug — commonly around 95mg, but genuinely variable with bean, grind and brew strength.
- Single espresso — around 60–80mg. Counter-intuitively less than a mug of filter, because the serving is tiny even though it is concentrated.
- Instant coffee — typically 60–80mg per teaspoon.
- Black tea — roughly 40–50mg; green tea lower again.
- Energy drinks — 80mg in a small can, 160mg in a large one, and considerably more in the concentrated ones.
- Cola — around 35mg per can.
- Pre-workout supplements — frequently 150–300mg per scoop, and the most commonly underestimated source on this list.
Two practical consequences. Chain coffee-shop drinks are often much larger than a home mug, so a "coffee" can be two or three doses by this reckoning. And decaf is not caffeine-free — usually a few milligrams, which is irrelevant for a single cup and can matter if you drink several late.
Working backwards from bedtime
The calculator gives you the hours a dose needs; the useful move is to invert your day around it. If you want to be asleep at eleven and your usual afternoon drink is a 160mg can, you need 8.4 hours — so the honest deadline is about half past two, not five o'clock.
People are often surprised by how early that lands, and the temptation is to conclude the model is too strict. Before doing that, it is worth testing it for a week: hold to the computed deadline, see whether sleep changes, and adjust the threshold or the half-life to match what you observe. If nothing changes, caffeine was not your constraint and you have learned something more useful than a rule.
Switching, rather than stopping
The practical middle ground for most people is not abstinence but substitution at the deadline. Tea instead of coffee roughly halves the dose; decaf removes almost all of it; and simply having the same drink two hours earlier is often enough on its own, because two hours is close to half a half-life and the decay is proportional.
That last point is worth holding onto. Because the clearance is exponential, moving a drink earlier buys more than it feels like it should — and the closer to bedtime a dose lands, the more each hour of delay is worth.