When should you stop drinking coffee? The 8-hour caffeine cutoff explained
Caffeine has a 5 to 6 hour half-life, so a 3 PM coffee leaves you with a quarter of the dose at midnight. Here is the science of the cutoff time, and how to set yours.
Most people make the same mistake with coffee: they believe their cutoff time is whenever they "stop feeling it." The pharmacology says otherwise.
Caffeine does not stay in your bloodstream until you feel it leave. It stays in your bloodstream on a clean exponential decay, and the half it has not cleared by bedtime is more than enough to fragment your sleep. Your brain can compensate enough to feel sleepy. It cannot compensate enough to sleep deeply.
Here is the math.
Half-life: the rule that decides cutoff
Caffeine has a biological half-life of 5 to 6 hours in healthy adults. That means whatever dose you drank, half of it is still in your body 5 to 6 hours later. Half of that is still there 5 to 6 hours after that. And so on.
For a typical 200 mg coffee at 8 AM:
| Time | Caffeine remaining |
|---|---|
| 8 AM | 200 mg |
| 2 PM (6h) | 100 mg |
| 8 PM (12h) | 50 mg |
| 2 AM (18h) | 25 mg |
50 mg at bedtime is enough to reduce slow-wave sleep by a measurable amount. That is the deepest stage of sleep, the one that drives physical recovery and metabolic repair.
The 6-hour controlled trial
The cleanest evidence on this is Drake 2013, Journal of Clinical Sleep Medicine. Twelve healthy adults, in their own beds, took 400 mg of caffeine (about 4 cups of coffee) at three different times: 0, 3, or 6 hours before bed.
The results were brutal:
- 6 hours before bed: Total sleep time reduced by 1 hour. Subjective sleep quality dropped sharply.
- 3 hours before bed: Total sleep time reduced by 1 hour. Sleep onset (time to fall asleep) increased by ~30 minutes.
- At bedtime: Total sleep reduced by 1 hour 20 minutes. Sleep onset doubled.
The most important finding: subjects could not tell the difference between conditions in their own self-report on most nights. They thought they slept normally. Polysomnography showed they did not.
This is why "coffee doesn't affect my sleep" is one of the most common false beliefs in nutrition. People are comparing their already-disrupted sleep to itself.
Setting your personal cutoff
Three factors determine where your cutoff should land.
1. Your bedtime
The 8-hour rule is a reasonable default. For a:
- 10 PM bedtime → last caffeine at 2 PM
- 11 PM bedtime → last caffeine at 3 PM
- Midnight bedtime → last caffeine at 4 PM
If you have any sleep complaints (long sleep onset, mid-night wake-ups, early-morning wake-ups), tighten that to 10 to 12 hours and last coffee before noon for two weeks. The improvement is usually obvious by week two.
2. Your dose
A 400 mg dose at 4 PM is much worse than a 80 mg dose at 4 PM. Most American filter coffee runs 100 to 150 mg per 240 ml cup. Espresso runs 60 to 80 mg per shot. Cold brew can hit 200+ mg per cup. Energy drinks vary wildly (80 to 300 mg per can; check the label).
A practical rule: if you are within 8 hours of bed, the dose has to come down. Switch to half-caff, single shot, or decaf.
3. Your genetics
The CYP1A2 gene controls caffeine metabolism in the liver. About 10% of adults carry the fast variant (clear caffeine in 3 to 4 hours) and 40% carry the slow variant (8 to 12 hours). The remaining 50% are intermediate.
You cannot easily test this without a genetic panel, but you can infer it. If a single coffee at 2 PM keeps you up, you are slow. If you can drink coffee at 9 PM and sleep, you are fast. Most people who think they are fast are intermediate and not noticing the cost.
Pregnant women clear caffeine far slower (half-life can stretch to 15 hours in late pregnancy). Smokers clear it faster. Oral contraceptive users clear it slower. These are individually large effects.
What about pre-workout?
If you train in the late afternoon or evening, the cutoff math collides with the performance benefit of caffeine before exercise.
Two practical solutions:
- Train earlier in the day if possible. 7 AM to 4 PM training pairs cleanly with morning caffeine.
- Use a much smaller dose for evening training. 100 mg of caffeine produces 80% of the performance benefit of 300 mg, with a fraction of the residual at bedtime. Skip the high-dose pre-workout supplements after 4 PM.
What about tea, matcha, and chocolate?
All count toward your daily caffeine. Approximate doses per typical serving:
| Drink | Caffeine |
|---|---|
| Filter coffee, 240 ml | 100 to 150 mg |
| Espresso, single shot | 60 to 80 mg |
| Cold brew, 360 ml | 150 to 250 mg |
| Black tea, 240 ml | 40 to 60 mg |
| Green tea, 240 ml | 25 to 45 mg |
| Matcha, 1 g powder | 35 mg |
| Yerba mate, 240 ml | 65 to 130 mg |
| Dark chocolate, 30 g | 12 to 25 mg |
| Energy drink, 350 ml | 80 to 300 mg |
| Decaf coffee, 240 ml | 2 to 15 mg |
A 6 PM black tea is roughly 50 mg of caffeine. Half of it is still in your system at midnight. Same logic, smaller dose, same cutoff principle.
Caffeine and weight loss specifically
Caffeine increases resting metabolic rate by about 3 to 11% for several hours after a 100 to 200 mg dose. The effect is real but small (50 to 100 kcal/day for a regular drinker). The bigger effect on weight loss is what caffeine does to the next night's sleep, which feeds the cycle described in how sleep affects weight loss: worse sleep, more hunger the next day, +250 to 385 kcal of intake. The metabolism boost from afternoon coffee is dwarfed by the calorie cost of the bad sleep it produces.
There is also a tolerance issue. Daily caffeine drinkers see most of the metabolic effect blunted within 1 to 2 weeks. The boost is real for occasional users, mostly absent for habitual drinkers. The sleep cost is durable.
What to do this week
If you suspect caffeine is hurting your sleep:
- Move last caffeine before noon for 14 days. This is the cleanest test. Most people who try it feel a difference within 5 to 7 nights.
- Cap total daily caffeine at 300 mg. Roughly 2 to 3 cups of coffee. Above this, the sleep cost rises faster than the alertness benefit.
- Switch evening drinks to decaf or herbal. Decaf coffee, peppermint, chamomile, and rooibos are all caffeine-free or near zero. Genuine coffee, tea, and sparkling water all hydrate the same; the only thing decaf gives up is caffeine.
- If you are caffeine-dependent, taper. Cutting 400 mg/day to zero overnight produces a 7 to 10 day headache for most people. Drop 50 to 100 mg every 3 to 4 days instead.
What not to do
- Do not switch to "energy" supplements with non-caffeine stimulants (yohimbine, synephrine, etc) to dodge the cutoff. The sleep effect is similar or worse, and the safety profile is much weaker.
- Do not assume tolerance fixes the problem. Tolerance reduces the subjective feeling of caffeine. The sleep architecture effect persists in long-term users.
- Do not stack caffeine with alcohol thinking they cancel out. They do not. Alcohol disrupts sleep through one mechanism, caffeine through another. The two together produce worse sleep than either alone.
Bottom line
Caffeine works by blocking the sleep signal. Once you accept that, the cutoff math is simple. For most adults, last caffeine 8 hours before bed; for sensitive sleepers, before noon, regardless of bedtime.
The 3 PM coffee feels free. It is not. It is a loan against tomorrow's sleep, and the interest rate, in calories overeaten, is higher than the boost is worth.
Pairs with the rest of the Sleep & weight cluster: how sleep affects weight loss, and late-night eating and sleep quality. For the broader hydration story including coffee, tea, and sparkling water: do coffee and tea count toward water intake.