The termination of obstructive respiratory events is typically associated with arousal from sleep. The ventilatory response to arousal may be an important determinant of subsequent respiratory stability/instability and therefore may be involved in perpetuating obstructive respiratory events. In healthy subjects arousal is associated with brief hyperventilation followed by more prolonged hypoventilation on return to sleep. This study was designed to assess whether elevated sleeping upper airway resistance (R(UA)) alters the ventilatory response to arousal and subsequent breathing on return to sleep in patients with obstructive sleep apnoea (OSA). Inspired minute ventilation (V(I)), R(UA) and end-tidal CO(2) pressure (P(ET,CO(2))) were measured in 22 patients (11 men, 11 women) with OSA (mean +/-s.e.m., apnoea-hypopnoea index (AHI) 48.9 +/- 5.9 events h(-1)) during non-rapid eye movement (NREM) sleep with low R(UA) (2.8 +/- 0.3 cmH(2)O l(-1) s; optimal continuous positive airway pressure (CPAP) = 11.3 +/- 0.7 cmH(2)O) and with elevated R(UA) (17.6 +/- 2.8 cmH(2)O l(-1) s; sub-optimal CPAP = 8.4 +/- 0.8 cmH(2)O). A single observer, unaware of respiratory data, identified spontaneous and tone-induced arousals of 3-15 s duration preceded and followed by stable NREM sleep. V(I) was compared between CPAP levels before and after spontaneous arousal in 16 subjects with tone-induced arousals in both conditions. During stable NREM sleep at sub-optimal CPAP, P(ET,CO(2)) was mildly elevated (43.5 +/- 0.8 versus 42.5 +/- 0.8 Torr). However, baseline V(I) (7.8 +/- 0.3 versus 8.0 +/- 0.3 l min(-1)) was unchanged between CPAP conditions. For the first three breaths following arousal, V(I) was higher for sub-optimal than optimal CPAP (first breath: 11.2 +/- 0.9 versus 9.3 +/- 0.6 l min(-1)). The magnitude of hypoventilation on return to sleep was not affected by the level of CPAP and both obstructive and central respiratory events were rare following arousal. Similar results occurred after tone-induced arousals which led to larger responses than spontaneous arousals. V(I) for the first breath following arousal under optimal CPAP was greater in men than women (11.0 +/- 0.4 versus 7.6 +/- 0.6 l min(-1)). These results demonstrate that the ventilatory response to arousal is influenced by pre-arousal airway resistance and gender. Whether this contributes to the perpetuation of respiratory events and the pathogenesis of OSA is unclear.