SummaryPerformance on a simple addition task was measured during three schedules of frequent sleep disruption for 2 nights. Five young adults had their sleep briefly disturbed for 2 nights in 3 separate weeks either every 1 min, every 10 min, or at sleep onset after an undisrupted 2.5-h sleep period. Subjects were required to perform a two-number, two-digit addition problem as rapidly as possible on awakening. Main effects were found for sleep disruption condition and time of night, and a significant interaction between the two was also observed. Latency to response was longest for the 10-min condition on night 1, on night 2, however, response latencies were longest in the 1-min condition. Response latencies were fastest in the 2.5-h condition for both nights of disruption. Arousal thresholds were also gathered across both nights. Arousal thresholds were consistently the highest in the 1- and 10-min conditions for both nights of disruption, reaching maximum threshold levels at the end of night 1. Arousal threshold was significantly positively correlated with response latency. Sleep stages (slow-wave sleep (SWS), SWS + REM (SWSR), and total sleep time minus stage 1 sleep) were poor predictors of performance changes across the 2 disruption nights. The data were best explained by sleep continuity theory, which posits that a period of at least 10 min of uninterrupted sleep is required for restoration to take place.