Caffeine and Sleep: How Late Is Too Late for Your Last Cup?

“I can drink coffee right before bed and sleep fine” is one of the most common claims in the caffeine-and-sleep conversation, and for some people it may even be roughly true. The research complicates that claim in an important way: caffeine’s effect on sleep architecture can show up even when falling asleep itself feels unaffected, which means “I sleep fine after coffee” is measuring a different thing than what sleep labs actually find.

Caffeine’s half-life is longer than most people think

Caffeine has an average half-life in healthy adults of around five hours, with a wide individual range (roughly 2 to 10 hours in pharmacokinetic studies), meaning about half the caffeine from a cup consumed at, say, 4 p.m. is still circulating in your system around 9-10 p.m., and a meaningful amount remains even later than that. This is the basis for the commonly cited six-hour cutoff recommendation, though it is an average, not a fixed number — actual half-life varies significantly between individuals based on genetics, age, liver function, pregnancy (by late pregnancy the half-life can roughly triple, reaching 10-15 hours), and interactions with certain medications and hormonal contraceptives.

This matters because the caffeine consumed at what feels like a reasonable mid-afternoon time is often still active in the body well into the evening, working against sleep onset and sleep quality in ways that are easy to underestimate if you’re only counting “hours since my last cup” casually rather than accounting for how slowly it actually clears.

What caffeine does to sleep beyond just delaying sleep onset

Caffeine primarily works by blocking adenosine receptors in the brain — adenosine is a compound that builds up across the day and promotes the pressure to sleep, and caffeine’s stimulant effect largely comes from interfering with that signal. Beyond the more obvious effect of making it harder to fall asleep, controlled sleep-lab studies have found caffeine, even when consumed hours before bedtime, can:

• Reduce total sleep time, sometimes by a meaningful margin, even in people who do not report difficulty falling asleep.
• Reduce slow-wave (deep) sleep, the sleep stage most associated with physical restoration, even in some studies where subjective sleep quality was rated as only mildly affected.
• Increase the number of brief awakenings across the night, similar to the fragmentation pattern seen with alcohol, though through a different mechanism.

One frequently cited controlled study (Drake and colleagues, 2013) gave healthy adults 400 mg of caffeine — roughly the amount in four cups of brewed coffee — at bedtime, three hours before, or six hours before. Even at six hours, objectively measured total sleep time fell by more than an hour compared with placebo, though the study was small (12 people) and used a high dose. A 2023 meta-analysis pooling 24 studies found caffeine cut total sleep time by about 45 minutes and sleep efficiency by about 7% on average, and estimated that a standard cup of coffee (about 107 mg) should be consumed at least 8.8 hours before bed to avoid reducing sleep time. Both are part of why sleep researchers generally recommend a wider buffer than “a cup of coffee a couple hours before bed should be fine.”

Why “I sleep fine after coffee” can still be technically wrong

There is a real gap between subjective sleep experience and what objective sleep measurement shows. Someone who falls asleep at a normal time and does not remember waking up may still be getting measurably less deep sleep or more fragmented sleep than they would without the caffeine — effects that show up in sleep-lab data more reliably than they show up in how rested someone feels reporting the next morning. This does not mean subjective experience is meaningless, but it is a reminder that “I don’t notice a problem” and “there is no effect happening” are not the same claim.

Individual variability is genuinely large

Not everyone is affected equally, and this is not just wishful thinking — genetic variation in the liver enzyme (CYP1A2) primarily responsible for metabolizing caffeine produces real differences in how fast people clear it, commonly described as “fast” and “slow” metabolizer profiles. Regular caffeine users also develop some tolerance to certain effects (like the subjective jitteriness) though tolerance to sleep-architecture effects appears to be less complete than tolerance to the alertness effect, based on available research. Sensitivity can also change over the lifespan — sleep generally becomes lighter and more easily disrupted with age — so a cutoff that worked fine at 25 may need revisiting later, even though studies have not found a consistent age-related slowing of caffeine clearance in healthy adults.

A practical approach

• Use at least a six-hour buffer before bedtime as a starting default — and closer to eight or nine hours for a strong coffee or energy drink, per the 2023 meta-analysis — and adjust based on your own observed sensitivity — if you’re a known “slow” responder or notice sleep quality issues even with an earlier cutoff, move it earlier still.
• Don’t rely solely on how easily you fall asleep as your test. If you have access to any sleep-tracking data (understood with its own limitations) or simply pay attention to how rested you feel and how fragmented your sleep seems, that is a better signal than sleep-onset speed alone.
• Account for all caffeine sources, not just coffee — tea, energy drinks, some sodas, chocolate, and certain medications all contribute, and an afternoon “just one more” from an unexpected source can push the cumulative timing later than intended.
• Be aware that pregnancy, certain medications, and liver conditions can substantially slow caffeine clearance — if any of these apply to you, a more conservative cutoff, and a conversation with your clinician about appropriate caffeine intake generally, is reasonable. During pregnancy, the American College of Obstetricians and Gynecologists advises keeping total caffeine under 200 mg a day.

The honest takeaway

Caffeine’s roughly five-hour average half-life (longer for many people) means a meaningful amount is often still active well into the evening after what feels like a reasonably timed cup, and its effects on sleep architecture — reduced deep sleep, more fragmentation — can occur even in people who do not notice trouble falling asleep. “I sleep fine” based on subjective experience alone is measuring something real, but not the same thing sleep-lab research measures, and the honest evidence points toward a wider buffer before bed than casual intuition usually assumes.

References

• Drake, C., et al. (2013). Caffeine effects on sleep taken zero, three, or six hours before bedtime (400 mg dose, 12 participants): objectively measured sleep was reduced by more than an hour even at the six-hour timepoint. Journal of Clinical Sleep Medicine. PubMed
• Gardiner, C., et al. (2023). Meta-analysis of 24 studies found caffeine reduced total sleep time by about 45 minutes on average, and estimated a standard cup of coffee should be consumed at least 8.8 hours before bed to avoid reducing sleep time. Sleep Medicine Reviews. PubMed
• A pharmacokinetic comparison of young and elderly adults found no consistent age-related slowing of caffeine clearance; half-lives ranged from about 2.3 to 9.9 hours. PubMed
• American College of Obstetricians and Gynecologists. (2010). Moderate caffeine consumption during pregnancy — advises keeping total caffeine intake under 200 mg/day. Committee Opinion No. 462. ACOG

Related reading

• The 4-7-8 Breathing Technique (and Other Ways to Fall Asleep Faster)
• Napping Done Right: How Long, How Late, and Why Timing Matters
• Sleep Debt: Can You Actually ‘Catch Up’ on the Weekend?

This article is for general educational purposes and is not a substitute for personalized medical advice.