One 90 mg Coffee Needs 2.5 to 6.8 Hours Before Bed to Drop Under 50 mg. Your Half-Life Decides Which

The usual advice is a fixed cutoff: no coffee after 2 pm, or nothing within six hours of bed. The six-hour figure comes from a real study. In 2013 Christopher Drake’s team gave 12 healthy sleepers 400 mg of caffeine at bedtime, three hours before, or six hours before; all three cut objectively measured sleep by 1.1 to 1.2 hours, and the authors wrote that this supports avoiding substantial caffeine for at least six hours before bed.

But the time caffeine takes to clear varies a lot. The Institute of Medicine’s 2001 report on caffeine gives a mean half-life of about 5 hours in healthy adults, and a range of 1.5 to 9.5 hours. We put a single filter coffee (90 mg, EFSA’s typical value for 200 ml) and a double (180 mg) into the Caffeine Cutoff calculator at five half-lives to see how far back the cutoff really needs to be.

Hours before bed to drop under the line

Each cell is how long before bedtime the drink has to be, and the latest clock time for an 11 pm bedtime:

The 50 mg and 25 mg lines are not thresholds from a study. They are there to show the shape of the answer: no single line is agreed on. EFSA’s 100 mg is a single dose that “may” affect sleep in some adults, and Drake’s six-hour dose left roughly 174 mg at bedtime.

Two things fall out of the table. First, the six-hour rule is about right for an average or slow metaboliser trying to get one coffee under 50 mg, and generous for a fast one. Second, doubling the dose doesn’t double the wait: it adds exactly one half-life, because the coffee has to halve one more time. The two right-hand columns are the two left-hand ones shifted by one half-life.

The math, in one line

Caffeine left after t hours is dose × 0.5^(t ÷ half-life). Solving for the time to fall under a line gives half-life × log₂(dose ÷ line). One coffee from 90 mg to 50 mg is log₂(1.8) = 0.85 half-lives; to 25 mg it is log₂(3.6) = 1.85. That is the whole table: each row multiplies those numbers by its half-life.

The calculator applies the same formula to every drink in your day and adds them up, so a morning coffee still counts a little at bedtime. Two 90 mg coffees at 8 am and 3 pm leave 41 mg at 11 pm at a 5-hour half-life, and 70 mg at 8 hours.

Where this doesn’t apply

• You don’t know your half-life. There’s no simple home test. The practical version: if a 3 pm coffee keeps you up and it doesn’t bother someone else, you’re probably toward the slow end. Try the 8-hour row first.
• Pregnancy and medication. Caffeine clears much more slowly in pregnancy, EFSA’s limit is 200 mg a day rather than 400, and some medications change clearance too. A doctor or pharmacist is the right source here.
• Bigger cups. The table uses EFSA’s 200 ml filter coffee. Coffee-shop sizes and cold brew can be much stronger; check the menu and use the 180 mg columns or the calculator’s custom amount.
• Sleep problems with other causes. Falling under a line doesn’t guarantee good sleep, and caffeine is only one factor. Late screen time and an irregular schedule matter too.
• Feeling fine isn’t evidence. Drake’s participants mostly didn’t notice the sleep they lost, which is why a measured number is more useful than how you feel.

What to actually do

1. Pick the row that matches you: 5 hours if you have no idea, 8 hours if afternoon coffee seems to keep you awake.
2. Pick a line. 50 mg is a reasonable start; drop to 25 mg if you’re sensitive.
3. Read off the latest time for your usual drink, or enter your real day in the calculator.
4. Keep that cutoff for two weeks without changing anything else, so you can tell whether it made a difference.
5. If you also want to pick a bedtime that ends on a full cycle, the 90-minute cycle post covers the wake-time side.

Find your own caffeine cutoff →