The 24-hour sleep/wake distributions of untreated patients with narcolepsy-cataplexy and matched normal habitual nappers were compared using home ambulatory monitoring. Subjects followed their usual sleep patterns including, for the habitual nappers, a self-selected daytime nap. There were no differences in 24-hour totals of sleep between groups other than a small increase in SWS in narcolepsy. Narcolepsy showed greater amounts of day sleep (stages 2, SWS, REM and total sleep) and less night sleep (stage 2, total sleep). Data were collapsed into 5 min epochs and entered into a matrix. The data in the two groups were then "wrapped" (re-aligned) around the 24 hours with phase 0 as each of the times of: evening sleep onset, onset of SWS, mid-point of night sleep and moment of morning awakening. In habitual nappers alignment beginning at morning wake-up produced the highest amplitude, least temporal dispersion and greatest kurtosis of daytime sleep (naps). The 24-hour sleep/wake distribution curves of both subject groups (data aligned at morning wake-up) based on collapsed data into 5 min bins then underwent curve fitting using 15th order polynomial regression. As with visual analyses of the raw data, the curve fits confirmed that the peak in daytime sleep propensity in narcoleptics was earlier by about 40 (2.66 hours). It was concluded that decreased daytime amplitude of a circadian arousal system was the most parsimonious explanation for the increased amount, broader temporal distribution and relative phase advance of day sleep in narcolepsy and that, as well, such a mechanism could explain a number of other features of the disease.