The effect of end-expiratory occlusion on respiratory muscle activity was studied in 10 unsedated preterm infants during sleep. Electromyograms (EMG) of the upper airway were recorded from surface electrodes placed over the submental (SM) area; diaphragm (DIA) EMGs were obtained with identical electrodes over the right subcostal margin. Phasic SM EMG accompanied 56 +/- 36% of breaths during spontaneous breathing and increased to 80 +/- 26% (P less than 0.05) on the first inspiratory effort after occlusion. Occlusion increased peak amplitude (P less than 0.001) and total duration (P less than 0.005) of the SM EMG without significant changes in its initial rate of rise. In contrast, only the total duration of the DIA EMG increased (P less than 0.005) during occlusion. Inspiratory time increased from 470 +/- 120 to 720 +/- 210 ms (P less than 0.001) during the first occluded effort, but expiratory time did not change. With sustained occlusion, peak amplitude of the SM EMG progressively increased, but DIA EMG only significantly increased by the third occluded effort. Pharyngeal patency was invariably maintained throughout the induced airway occlusions. Sharp bursts of SM EMG activity coincided with resolution of spontaneous obstructive apneic episodes in four infants. The immediate increase in SM EMG associated with airway occlusion may be a mechanism that prevents the development of obstructive apnea.