Excessive daytime sleepiness is a complaint characterizing many disorders of the wakefulness--sleep cycle. This paper addresses the complaint of sleepiness objectively by an attempt to differentiate a group of control subjects from a group of patients with unambiguous narcolepsy. Fourteen control and 27 narcoleptic subjects were evaluated by one of three protocols involving nocturnal recordings, detailed interviews, and 5 or more 20-min opportunities to sleep offered at 2-h intervals beginning at 10.00 o'clock, +/- 30 min. Each 20-min opportunity to sleep was given to subjects lying in a darkened quiet room and asked to try to fall asleep. Polysomnographic variables were monitored and sleep was scored in 30-sec epochs by standard criteria. The interval from the start of each test to the first epoch of NREM (including stage 1 sleep) or REM sleep was called sleep latency. In two of the protocols, the subjects were awakened immediately after sleep onset. In the third protocol, the subjects were awakened after 10 min of sleep. Narcoleptics consistently fell asleep much more readily than did control subjects. We conclude that the Multiple Sleep latency test, in addition to providing opportunities to clinically document sleep onset REM sleep periods, can demonstrate pathological sleepiness. Based on these data, we suggest that an average sleep latency less than 5 min be set as the minimum cutoff point for pathological sleepiness.