Prolonged wakefulness increases electroencephalogram (EEG) low-frequency activity (< 10 Hz) in waking and sleep, and reduces spindle frequency activity (approximately 12-16 Hz) in non-rapid-eye-movement (nonREM) sleep. These physiologic markers of enhanced sleep propensity reflect a sleep-wake-dependent process referred to as sleep homeostasis. We hypothesized that caffeine, an adenosine receptor antagonist, reduces the increase of sleep propensity during waking. To test this hypothesis, we compared the effects of caffeine and placebo on EEG power spectra during and after 40 h of wakefulness. A total of 12 young men underwent two periods of sleep deprivation. According to a randomized, double-blind, crossover design, they received two doses of caffeine (200 mg) or placebo after 11 and 23 h of wakefulness. Sleep propensity was estimated at 3-h intervals by measuring subjective sleepiness and EEG theta (5-8 Hz) activity, and polysomnographic recordings of baseline and recovery nights. Saliva caffeine concentration decreased from 15.7 micromol/l 16 h before the recovery night, to 1.8 micromol/l 1 h before the recovery night. Compared with placebo, caffeine reduced sleepiness and theta activity during wakefulness. Compared with sleep under baseline conditions, sleep deprivation increased 0.75-8.0 Hz activity and reduced spindle frequency activity in nonREM sleep of the recovery nights. Although caffeine approached undetectable saliva concentrations before recovery sleep, it significantly reduced EEG power in the 0.75-2.0 Hz band and enhanced power in the 11.25-20.0 Hz range relative to placebo. These findings suggest that caffeine attenuated the build-up of sleep propensity associated with wakefulness, and support an important role of adenosine and adenosine receptors in the homeostatic regulation of sleep.