Multiple sleep latency tests were performed in 42 drug-free outpatients with excessive sleepiness: 12 narcoleptics, 9 sleep apneics, 7 primary depressives, and 14 patients with miscellaneous disorders. Among-group comparisons were made by one-way ANOVA for each nap (time-of-day effect) and using four-nap means of EEG sleep variables. Four-nap means were significantly different among groups for percent awake, percent time spent asleep, stage 1 latency, and REM latency. Of 22 significant pairwise comparisons (P less than 0.05, LSD test), 77.3% occurred at 12.00 and 14.00. Depressives showed lower sleep percentage (more arousal) and fell asleep later than narcoleptics or apneics. Patients with miscellaneous disorders occupied a middle position. Sleep percentage was gradually reduced during the day in depressives but remained high or rose further in apneics, narcoleptics, and miscellaneous patients. Naps were SOREMP-positive (sleep onset REM) 60.4% of the time in narcoleptics, 25.0% in apneics, 3.5% in depressives, and 5.4% in miscellaneous patients. SOREMP distribution across naps (10.00, 12.00, 14.00 and 16.00) was 19%, 31%, 19% and 31%, respectively. REM latency was significantly shorter in narcolepsy than in apnea. In summary, results show a continuum of excessive daytime sleepiness (EDS), demonstrating MSLT application in the differential diagnosis of EDS and significant diurnal variation in sleep measures.