Twelve narcoleptic subjects experiencing at least five daytime REM sleep episodes were monitored for 34 consecutive hours in the laboratory starting at 2200 h one evening and ending at 0800 h a day and a half later. There was no significant difference between the length of the daytime and nighttime REM cycles. To test the hypothesis that an underlying rhythm governs REM episodes, a grid was constructed on the basis of the mean and the SD of the daytime cycles (starting at 0800 h the first morning) was projected on the following nighttime and its correspondence with actual night cycles was evaluated. Overall, it was observed that the number of night cycles falling within the projected grid was significantly higher than chance (p < 0.01), indicating that nighttime REM episodes tended to fall within the same periodicity as their preceding daytime episodes. This observation supports the hypothesis that an underlying basic rest-activity cycle governs REM sleep episodes in narcoleptic subjects.