Recently, we have shown that during the response to a more prolonged (20 min) hypoxia in preterm infants (Lemke, et al, Pediatr. Res., 39:389A, 1996), ventilation falls after initial hyperventilation and remains depressed between 5 and 20 min. To test the hypothesis that the response is influenced by the infant's sleep state and the presence or absence of arousal we studied 14 healthy preterm infants [BW 1.6±0.1 kg (mean±SE), SW 2.0±0.2 kg, GA 31±1 wk, PNA 27±4 d]. Ventilation was measured using a flow-through system. Sleep states were defined using EEG, EOG, and body movements. Arousal was defined using eye opening, body movements, agitation, and crying. The experimental procedure consisted in giving the infants 21% O2 for 2 min, 15% O2 for 20 min, followed by 2 min of recovery in 21% O2. Fourteen experiments were conducted in quiet sleep and 7 in REM sleep. Arousal was observed in 71% of experiments in quiet sleep (10 out of 14) and 0% in REM sleep (0 out of 7; p<0.01). In quiet sleep, minute ventilation remained above baseline at 5(6.7±2.5%), 10 (7.4±2.8%), 15 (3.4±2.1%), and 20(6.3±2.7%) minutes of hypoxia. On the contrary, in REM sleep minute ventilation decreased below baseline at 5 (-4.2±2.3), 10(-4.5±5.0), 15 (-7.4±2.3%), and 20 (-7.9±3.1%) minutes of hypoxia (p<0.001 quiet vs REM for all time points). During quiet sleep minute ventilation at 10, 15, and 20 min of hypoxia was significantly higher in infants who exhibited arousal (12.6±3.1%, 8.6±2.3%, and 12.0±3.1% respectively) than in infants who did not (-5.6±4.9%,-8.2±3.5%, and -4.8±4.4% respectively; p<0.001). The difference in minute ventilation at each time point was primarily related to a decrease in frequency with no significant change in VT. These observations suggest 1) preterm infant's ventilatory response to prolonged hypoxia is sleep state dependent, 2) arousal is more frequent during quiet sleep than in to REM sleep. We speculate that sleep state determines ventilatory and arousal responses to prolonged hypoxia in preterm infants which in turn may determine the infant's predisposition to SIDS.Supported by the Children's Hospital of Winnipeg Research Foundation.