Caffeine Half-Life: What Drake et al. Actually Found

Everyone has a caffeine cutoff. For most people, it is a vague internal rule — "no coffee after 2 pm," or "only one in the afternoon" — based on gut feeling. The problem with gut feeling is that it is a terrible sleep scientist. One study, more than any other, proved exactly how terrible.

1. The Study Everyone Cites but Few Have Read

In 2013, Christopher Drake and colleagues at the Henry Ford Hospital Sleep Disorders Center published a small, clean study in the Journal of Clinical Sleep Medicine. The design was simple: take 12 healthy adults who regularly drink caffeine, give them 400 mg (roughly the equivalent of two large coffees) at three different time points before bedtime, and see what happens.

The four conditions — caffeine at bedtime, 3 hours before, 6 hours before, and placebo — were run in a randomised, double-blind, crossover design. Each participant completed all four conditions in their own home, with sleep tracked by both a validated wireless EEG headband and a standard sleep diary.

If you have ever Googled "how long before bed should I stop drinking coffee," this study is almost certainly the one behind the answer. It has been cited thousands of times. But the headline stat most people pull — "6 hours is not enough" — actually undersells what the data showed.

2. The Numbers: 0, 3, and 6 Hours Before Bed

All three caffeine conditions significantly reduced total sleep time compared to placebo. Not just the bedtime dose. Not just the 3-hour dose. Caffeine consumed a full 6 hours before bed still cut total sleep by approximately 1 hour.

The 3-hour condition extended the time it took to fall asleep by 17 minutes and increased wakefulness during the night by nearly 28 minutes. The 6-hour condition added 8 minutes of wake time during sleep. Across all caffeine conditions, light sleep (stages 1 and 2) dropped by 40 to 44 minutes, and slow-wave deep sleep was reduced in both the bedtime and 6-hour conditions.

If you are curious how quickly you typically drift off once your head hits the pillow, the Sleep Latency Calculator quantifies that window — and a consistently short latency (under 5 minutes) can itself be a sign of accumulated sleep debt.

3. The Part Nobody Talks About

Here is the finding that should genuinely change behaviour: participants in the 6-hour condition had no idea their sleep was disrupted. The EEG headband registered the damage — over an hour of lost sleep, reduced deep sleep, increased wakefulness. But when those same participants filled in their sleep diaries the next morning, they reported sleeping fine.

This subjective-objective gap is why "I can drink coffee at 4 pm and sleep perfectly" is the most unreliable statement in sleep hygiene. You may fall asleep on time, lie unconscious for what feels like a full night, and wake up none the wiser. But the architecture of your sleep — the deep, restorative stages your brain needs — has been quietly degraded.

4. Half-Life Is Not One Number

The phrase "caffeine has a half-life of about 5 hours" appears in virtually every popular health article. It is a useful average and a terrible individual prediction.

A 2022 systematic analysis of 141 pharmacokinetic studies covering 4,714 individuals found that caffeine half-life ranges from approximately 2 to 10 hours across the general population. The primary driver is CYP1A2, the liver enzyme responsible for roughly 90 percent of caffeine metabolism. Its activity varies up to 6-fold between individuals.

At the fast end, someone with highly active CYP1A2 could drink a coffee at 3 pm and have less than 12 percent of the dose circulating by 9 pm. At the slow end, the same coffee at the same time could still be nearly 50 percent active at midnight. Same cup. Same clock. Radically different outcome.

Smoking significantly increases CYP1A2 activity, which shortens caffeine half-life. Oral contraceptives reduce it, which extends half-life. And twin studies indicate that when you control for those environmental factors, 89 percent of the variation in caffeine clearance is genetically determined.

Our Caffeine Cutoff Calculator accounts for this variation — enter your bedtime and it gives you a personalised cutoff with a live caffeine decay chart, so you can see exactly when the dose drops below the threshold that matters.

5. Your Genes Decide Your Sensitivity

Even if two people metabolise caffeine at the same rate, they can still react to it very differently. The reason sits in a different gene entirely: ADORA2A, which codes for the adenosine A2A receptor — the very receptor caffeine blocks.

A 2007 study by Rétey and colleagues identified a variation in ADORA2A that determines individual caffeine sensitivity. Carriers of the C/C genotype showed EEG patterns resembling insomnia after caffeine — increased high-frequency brain oscillations that signal lighter, less restorative sleep. The effect was halved in C/T carriers and absent in T/T carriers.

In a follow-up survey, roughly 31 percent of adults self-identified as caffeine-sensitive. A large 2019 replication study by Erblang and colleagues in 1,023 French workers confirmed the genetic link: ADORA2A variants predicted both sleep complaints and insomnia risk. Intriguingly, the genetic differences only appeared in low caffeine consumers (under 50 mg per day). At high daily intakes (above 300 mg), the genetic advantage disappeared — everyone's sleep suffered.

If you already know caffeine hits you harder than most, try the Chronotype Quiz as well — evening chronotypes tend to report greater caffeine sensitivity, and misaligning your caffeine window with your circadian rhythm compounds the damage.

6. Regular Use Does Not Build Full Tolerance

A common defence: "I drink 3 coffees a day — I've built tolerance." The evidence says: not entirely.

A 2024 randomised crossover trial by Gardiner and colleagues tested multiple dose-timing combinations in 23 healthy habitual caffeine consumers. At 400 mg consumed 4 hours before bed, total sleep time dropped by 50 minutes, deep sleep by nearly 30 minutes, and perceived sleep quality fell by 34 percent — in people who drank caffeine daily. At 8 hours out, sleep efficiency was still reduced by 7 percent. Even at 12 hours, the 400 mg dose left a measurable footprint.

Separately, Weibel and colleagues (2021) tracked 20 habitual caffeine consumers through 10-day treatment periods and found that regular caffeine use delayed REM sleep onset by 25 minutes — 78.7 minutes to reach the first REM episode, versus 53.6 on placebo. REM proportion was also significantly reduced.

REM sleep is where your brain consolidates emotional memories, processes learning, and performs a kind of overnight therapy. Shortchanging it does not produce obvious daytime drowsiness the way losing deep sleep does. It shows up more subtly — as irritability, difficulty with creative problem-solving, and impaired emotional regulation.

If you have ever wondered whether your sleep quality score is being quietly undermined by a habit you have had for years, this is the mechanism. The Sleep Cycle Calculator can help you structure your night around full 90-minute cycles, giving REM the space it needs in the later cycles.

7. What the Meta-Analysis Actually Recommends

Individual studies are useful. A meta-analysis pooling 24 of them is more useful.

Gardiner and colleagues' 2023 systematic review in Sleep Medicine Reviews found that, on average, caffeine reduced total sleep time by 45 minutes, increased sleep onset latency by 9 minutes, decreased sleep efficiency by 7 percent, and cut deep sleep by 11.4 minutes. For every additional milligram of caffeine consumed, total sleep loss increased by 0.2 minutes. For every additional hour between consumption and bedtime, sleep loss decreased by 2.8 minutes.

From that dose-timing model, the authors derived specific recommendations: at least 8.8 hours before bed for a standard cup of coffee (107 mg), and at least 13.2 hours for a pre-workout supplement (217.5 mg). Those are evidence-based numbers, not vibes.

If that 8.8-hour cutoff lands uncomfortably early for your schedule, it may be worth examining whether your daytime energy dip is a sleep debt problem rather than a caffeine problem — and whether an earlier exercise window could replace the afternoon cup entirely.

The Bottom Line

Drake et al. did not discover that caffeine is bad for sleep. We already knew that. What they showed — in a clean, controlled design — is that the disruption reaches further than most people assume, and that we cannot feel it happening. Six hours is not a safe margin. Your body is not an exception to the pharmacokinetics.

The real number depends on your genetics (CYP1A2 for clearance speed, ADORA2A for receptor sensitivity), your dose, and your personal sleep architecture. But the meta-analytic evidence points to a hard minimum of about 9 hours for an average cup of coffee. If your cutoff is "no coffee after 2 pm" and your bedtime is 10:30, the maths checks out. If your cutoff is vaguer than that, it probably doesn't.

Not vibes. Real research.

This article is for informational purposes only and does not constitute medical advice. If you have persistent sleep difficulties, please consult a qualified healthcare provider.