Thirty-six healthy young men and women (age range 21-35 years) were studied in an experimental model of sleep fragmentation. On 2 nights sleep was disrupted by presenting tones to produce brief electroencephalogram (EEG) arousals (without shortening sleep time) and daytime function was assessed the following day with the Multiple Sleep Latency Test and a divided attention performance test. The fragmentation of sleep produced significant disruption of nocturnal sleep and reduced daytime alertness. Adaptation in EEG-defined arousals occurred from the 1st to the 2nd night of fragmentation. Threshold (measured indirectly) characteristics of EEG-defined arousals were somewhat different than those of previous studies requiring behavioral awakening. The percent of tone series producing arousal, number of tones necessary for arousal and duration of the arousal all reflected heightened thresholds in stage 3/4 and rapid eye movement (REM) sleep compared to stage 1 and stage 2 sleep. In the last 3 hours of sleep versus the first 3 hours, arousals occurred less frequently, required more tones to produce, resulted in shorter durations and in fewer sleep stage changes, except for REM sleep where the converse was the case.