Following a baseline night recording, 8 narcoleptic subjects and 8 sex- and age-matched controls were maintained on a 90 min sleep/wake schedule for 48-72 h. Each cycle consisted of 60 min of enforced wakefulness out of bed, followed by a 30 min "nap" period in which subjects were asked to try and fall asleep. Upon completion of the 90 min sleep/wake protocol, subjects were permitted to sleep ad libitum for 24 h. All sleep periods were monitored polygraphically; in addition, tympanic temperature and subjective sleepiness were recorded during the 90 min sleep/wake schedule. Narcoleptics and controls differed dramatically in their sleep patterns during the 90 min sleep/wake schedule. On average, narcoleptics obtained 2 more hours per day of total sleep time (TST) than did the controls, with REM sleep comprising nearly 2/3 of the incremental sleep time. The two groups did not differ with respect to the amount of slow wave sleep (stage 3 + 4; SWS). The sleep latency rhythms observed in control subjects were markedly diminished in narcoleptics; narcoleptic subjects remained objectively sleepy (i.e., had low sleep latencies) even at times corresponding to maximum alertness in the control subjects. Rhythms in subjective sleepiness and core temperature were, however, robust in both groups. Although TST in narcoleptics exceeded that of controls during the 90 min sleep/wake schedule, narcoleptics did not obtain more sleep than controls during the baseline or recovery periods. These findings suggest that the homeostatic process of sleep regulation is intact in narcoleptics. Moreover, it appears that the circadian clock itself is functioning normally in narcoleptics. An attenuated clock effector mechanism responsible for promoting alertness may, however, explain excessive daytime sleepiness in narcoleptics.