Eleven narcoleptic and 11 control subjects completed five multiple sleep latency test (MSLT) procedures with instructions to "fall asleep" while supine and five maintenance of wakefulness test (MWT) procedures with instructions to "remain awake" while comfortably sitting. Narcoleptic subjects had a shorter sleep latency and a higher frequency of sleep onset REM periods (SOREMPs) on both daytime tests than controls. Each group had a longer sleep latency on the MWT than MSLT. For patients with narcolepsy, the differences between the two daytime procedures suggest the tests measure distinct aspects of sleep-wake tendency. Sleep latency on the MSLT did not correlate with sleep latency on the MWT for narcoleptic subjects. Furthermore, the number of SOREMPs during the MSLT for narcoleptic subjects did not correlate with the number of SOREMP during the MWT. Measures of nocturnal REM sleep for narcoleptic subjects correlated with sleep latency on the MSLT. In particular, REM latency at night was highly predictive of the magnitude of hypersomnia for patients with narcolepsy. Sleep latency at night tended to relate to sleep latency on the daytime tests for controls. Patients with narcolepsy had a shorter sleep latency, more frequent arousals, and a shorter REM latency than control subjects across the 2 nights of study.