Caffeine and sleep: how late is too late?

“No coffee after 2 pm” is the most common caffeine rule, and it is not a bad one. But it treats a single espresso and a 300 mg energy drink the same, and it ignores the fact that some people clear caffeine twice as fast as others. The research now lets us do better.

What caffeine does at night

Caffeine works mainly by blocking adenosine, a messenger that builds up while you are awake and contributes to sleep pressure ¹. Blocking it does not remove your need for sleep. It hides the signal, which is why caffeine can leave you lying awake while still tired. Brain recordings show the effect directly: caffeine reduces the slow-wave activity that characterises deep sleep ².

Caffeine also shifts the body clock. In a placebo-controlled laboratory study, a dose equivalent to a double espresso three hours before bed delayed the melatonin rhythm by about 40 minutes ³. So a late coffee can make it harder to fall asleep tonight and nudge your clock later for tomorrow.

How long it stays in your system

Caffeine levels typically peak 30 to 120 minutes after you drink it, and its half-life is usually quoted at 3 to 6 hours ⁴. After one half-life, half the dose is still circulating; after two, a quarter.

Those averages hide a lot of variation. One liver enzyme, CYP1A2, handles about 95% of caffeine metabolism, and common genetic differences in it make clearance highly variable between people ⁵. Age, liver function, some medications and smoking all change it too. A systematic analysis of 141 pharmacokinetic studies confirmed that smoking speeds caffeine clearance while oral contraceptive use slows it ⁶. In one classic study, women using low-dose oestrogen-containing contraceptives had an average caffeine half-life of 7.9 hours, against 5.4 hours in women who were not ⁷.

What the trials show

The overall effect

The best summary comes from a 2023 meta-analysis of 24 studies ⁸. On average, caffeine:

An earlier systematic review reached the same conclusions and added two important caveats: older adults may be more sensitive, and most studies were done in young men in Western countries, which limits how far the findings generalise ².

Six hours is not always enough

In a placebo-controlled study, people took 400 mg of caffeine at bedtime, three hours before, or six hours before. All three timings significantly disturbed sleep compared with placebo, measured both by questionnaires and a home sleep monitor. The authors concluded that substantial caffeine should be avoided for at least six hours before bed ⁹.

400 mg is a large single dose, nearly twice the standard pre-workout serving used in the meta-analysis below, but some products and stacked drinks get there.

Dose changes everything

A 2025 randomised crossover trial tested two doses at three different times in 23 young men ¹⁰. Each took placebo, 100 mg or 400 mg of caffeine 12, 8 or 4 hours before bed, with in-home sleep recordings:

• 100 mg had no significant effect on objective or self-reported sleep at any timing, including 4 hours before bed.
• 400 mg delayed sleep onset and altered sleep architecture when taken within 12 hours of bed, and caused more fragmented sleep within 8 hours.

The study also found a gap between measured and perceived sleep: the recordings picked up effects that participants did not report, which the authors took to mean people struggle to judge how caffeine affects their sleep. “Caffeine doesn’t affect me” is not reliable evidence.

Even morning caffeine?

One small study gave nine healthy men 200 mg of caffeine at 7:10 am. That night their sleep efficiency and total sleep time were lower than after placebo, even though caffeine levels in their saliva had fallen to a low level by bedtime ¹. It is a single, small study, but it suggests the most sensitive people may be affected by morning caffeine too.

So when should you stop?

The 2023 meta-analysis translated its findings into cut-off times to avoid losing total sleep ⁸:

The 2025 trial suggests a smaller dose, around 100 mg, may be fine closer to bed for many people ¹⁰. These two findings fit together better than they first appear: the meta-analysis averaged many studies, many with larger doses, while the trial isolated dose and timing directly. Both point to the same principle. The bigger the dose, the earlier it needs to be.

A sensible approach for most people:

1. Keep large doses in the morning. Anything over about 200 mg in one go belongs early in the day.
2. Treat the afternoon as small-doses-only. One ordinary coffee in the early afternoon is probably fine for most people. A large energy drink is not.
3. Adjust for your biology. If you are older, use oestrogen-containing contraception, or simply know you are sensitive, move your cut-off earlier.
4. Test it on yourself. Try two weeks with your usual pattern and two weeks with an earlier cut-off, and compare how long you take to fall asleep and how rested you feel.

Our caffeine cut-off calculator applies these research figures to your own bedtime and drinks, and shows a range rather than a single number, because nobody’s metabolism is average.

Who is more sensitive?

Some people genuinely tolerate late coffee better than others. A variant in the gene for the adenosine A2A receptor, caffeine’s main target in the brain, was more common in people who described themselves as caffeine-sensitive, and it affected how strongly caffeine changed their sleep EEG ¹¹. Variation in the enzyme that metabolises caffeine adds a second layer ⁵.

That said, consumer genetic tests are not validated for setting personal caffeine cut-offs. Your own sleep is a better guide, provided you remember that you may not notice the effects without measuring.

Our take

Caffeine is one of the clearest cases in sleep science: the effect is real, it scales with dose and timing, and it is easy to underestimate. You do not need to give up coffee. Move the big doses early, keep the afternoon light, and if you are struggling with sleep, make caffeine timing the first thing you test.

If you take L-theanine alongside caffeine, see our L-theanine article for what it does and doesn’t do for sleep.