We hypothesized that the increased arousal threshold to upper airway occlusion exhibited by patients with obstructive sleep apnea (OSA) is in part secondary to the disease process itself. To test this hypothesis, we studied the effects of withdrawal of three nights of nasal continuous positive airway pressure (CPAP) treatment on arousal in six male patients with severe OSA who were using nasal CPAP on a long-term basis. During the control week, patients slept with nasal CPAP at home and on the first of 2 nights in the sleep laboratory (night C1, CPAP; night C2, no CPAP). During the apnea week, patients slept without nasal CPAP for 2 nights at home and 2 nights in the sleep laboratory (AP1, AP2). The control and apnea weeks were consecutive and in random order. The mean (+/-SEM) apnea+hypopnea index was 76.9 +/- 7.1 on AP1 vs 3.1 +/- 1.0 events per hour on C1 (p<0.05). Thus, the laboratory night (and presumably the 2 nights at home) preceding AP2 had dramatic increases in apnea compared with the nights preceding C2. The apnea duration during nonrapid eye movement sleep on nights following apnea was greater (AP2: 28.7 +/- 1.5 vs C2: 25.5 +/- 1. 7 s; p<0/05) and the arousal threshold as reflected by the maximum esophageal pressure deflection preceding arousal was higher (DPmax) (AP2: 55.1 +/- 5.7 vs C2: 45.3 +/- 6.4 cm H2O; p<0.005). We conclude that prior sleep apnea increases the arousal threshold to upper airway occlusion on subsequent nights and prolongs the apneic events.