Experiments were done on seven lambs to determine if site of occlusion--nasal versus tracheal--influences the cardiopulmonary and arousal responses from sleep to upper airway obstruction. Each lamb was anesthetized and instrumented for sleep staging and measurements of heart rate and arterial hemoglobin oxygen saturation. A tracheostomy was also done and a fenestrated tracheostomy tube placed in the trachea. Prior to an experiment, A 5F balloon-tipped catheter was inserted through the decannulation cannula into the tracheostomy tube so that tracheal occlusions could be accomplished by inflating the balloon. In addition, a 5F balloon-tipped catheter was inserted into the inlet of a pre-formed silicone mask sealed to the animals snout with silicone rubber foam so that nasal occlusions could be accomplished by inflating the balloon. During an experiment, measurements were made in quiet sleep and in active sleep during control periods of tidal breathing and during experimental periods of nasal or tracheal occlusion. Upper airway obstruction was terminated by deflating the balloon once the animal aroused from sleep. Arousal occurred sooner following nasal occlusion than during tracheal occlusion in quiet sleep; 64 percent of arousals occurred within five seconds of nasal occlusion whereas only 14 percent of arousals occurred within five seconds of tracheal occlusion in quiet sleep. In addition, SaO2 and heart rate decreased more before arousal following tracheal occlusion than following nasal occlusion. However, there was not a significant effect of site of obstruction on time to arousal or the change in SaO2 before arousal in active sleep.(ABSTRACT TRUNCATED AT 250 WORDS)