Sleep Science
Caffeine, adenosine and the reason coffee stops working
Caffeine does not supply energy. It blocks the signal that tells you how tired you are, and the debt it conceals is still accruing underneath.
Adenosine is a by-product of cellular energy metabolism. As neurons consume ATP through the day, adenosine accumulates in the extracellular space of the brain. Binding to its receptors — principally A1 and A2A — it inhibits arousal-promoting systems and promotes sleep.
It is the molecular substrate of what sleep researchers call homeostatic sleep pressure, or Process S: the longer you are awake, the more of it there is, and the harder it becomes to stay awake. During sleep, adenosine is cleared, and the pressure resets.
What caffeine actually does
Caffeine is structurally similar enough to adenosine to fit its receptors, and it binds without activating them. It is an antagonist — it occupies the parking space without doing the job.
The consequence is precise and worth stating carefully. Caffeine does not remove adenosine, reduce its production, or supply energy. It prevents you from perceiving the sleep pressure that has already built up. The adenosine keeps accumulating behind the blockade.
Which explains the crash. As caffeine is metabolised and clears the receptors, the accumulated adenosine — now at a higher concentration than when you drank the coffee — binds all at once. You do not merely return to how you felt; you feel worse.
Caffeine's half-life in healthy adults averages roughly five hours, with wide individual variation. A 200 mg coffee at 3pm leaves about 100 mg at 8pm and 50 mg at 1am — roughly half a cup still circulating as you try to sleep.
The quarter-life is the more useful concept: about ten hours to fall to a quarter. Which is why sleep researchers commonly advise no caffeine after early afternoon, and why people who insist coffee does not affect their sleep are frequently wrong about it.
Why people vary so much
Caffeine is metabolised primarily by the liver enzyme CYP1A2, and variation in the CYP1A2 gene produces fast and slow metabolisers. Slow metabolisers clear it over considerably longer, and are more likely to experience anxiety, palpitations and sleep disruption from a given dose.
Variation in the adenosine A2A receptor gene ADORA2A affects sensitivity independently of clearance, and is associated with both caffeine-induced anxiety and with sleep disturbance.
Other modifiers: smoking induces CYP1A2 and roughly halves caffeine half-life, which is why people who quit smoking often become suddenly caffeine-sensitive on their usual intake. Pregnancy substantially prolongs it — up to fifteen hours in the third trimester. Oral contraceptives roughly double it. Some medications, including certain antibiotics and antidepressants, inhibit the enzyme.
What it does to sleep architecture
Even where people fall asleep normally, caffeine measurably alters sleep. Controlled studies find reduced total sleep time, increased sleep latency, reduced slow-wave sleep and increased stage 1 and 2 — a shift toward lighter sleep.
A well-designed study by Christopher Drake and colleagues gave 400 mg of caffeine at 0, 3 and 6 hours before bed. All three conditions produced significant sleep disruption compared with placebo, including the dose taken six hours before bedtime. Notably, participants did not reliably perceive the disruption in the six-hour condition.
That last point is the crux. Self-report is not a reliable guide to whether caffeine is affecting your sleep.
Tolerance and withdrawal
Regular use produces upregulation of adenosine receptors — the brain compensates for the blockade by making more parking spaces. Tolerance to the alerting effects develops within days to weeks, though tolerance to the sleep-disrupting effects appears to be incomplete.
The consequence is that habitual coffee drinkers are, to a substantial degree, drinking to return to baseline rather than to exceed it. Withdrawal — headache, fatigue, irritability, poor concentration, low mood — begins twelve to twenty-four hours after the last dose, peaks at one to two days, and resolves over about a week.
Studies comparing habitual consumers deprived overnight against non-consumers find that much of the apparent benefit of morning coffee is the reversal of overnight withdrawal. Whether any net alerting benefit remains for chronic users is genuinely debated.
Using it well
Delay the first cup. Cortisol peaks naturally in the first hour after waking, and sleep pressure is at its lowest. Caffeine adds little there. Waiting an hour or two makes better use of it.
Set a cut-off. Ten hours before bedtime is conservative; eight is a reasonable compromise for most people. For a midnight sleeper, that means nothing after about 2pm to 4pm.
Know your actual dose. A filter coffee is roughly 95 mg, an espresso around 65, instant around 60, black tea around 45, green tea around 30, a can of cola around 35, energy drinks 80 and upward. Portion sizes in cafés have grown considerably.
The coffee nap. Drink coffee, immediately sleep for 20 minutes, wake as the caffeine arrives. It works because the nap clears some adenosine while the caffeine blocks the rest, and it has been tested with positive results on alertness and driving performance.
Do not use it to cover chronic sleep restriction. Caffeine masks impairment without removing it — a well-established finding in driving research, where caffeinated but sleep-deprived drivers feel considerably safer than they are.
The reassuring part
Moderate caffeine intake — up to around 400 mg a day in healthy adults — is regarded as safe by most regulatory bodies and is associated in cohort studies with neutral or slightly favourable outcomes for several conditions.
The issue is not that caffeine is dangerous. It is that it is a very effective way of concealing a sleep problem from yourself.
