Pharyngeal occlusion during obstructive apnea is thought to be an inspiratory-related event; however, occlusion also occurs in the absence of negative intrathoracic pressure. We hypothesized that inspiratory-related pharyngeal occlusion would be preceded by significant expiratory narrowing. Eight sleeping patients with obstructive apnea were studied. Pharyngeal caliber, airflow, and esophageal pressure (Pes) were simultaneously monitored during three to four consecutive breaths preceding occlusion (between 3 and 22 events were studied per subject). Relative changes in retropalatal airway cross-sectional area (CSA) were determined from fiberoptic images (five frames per second) normalized to the maximum CSA. During inspiration, CSA was significantly reduced only during the breath immediately preceding the apnea (Group mean CSA +/- SEM: 51 +/- 8% at the start of inspiration compared with 37 +/- 8% at midinspiration). During expiration, for all breaths there was an initial significant increase in CSA compared with the nadir CSA during the preceding inspiration (CSA: breath-3, 57 +/- 10% to 79 +/- 3%; breath-2, 59 +/- 8% to 76 +/- 4%; breath-1, 37 +/- 8% to 64 +/- 8%), followed by a significant narrowing at end-expiration compared with the peak CSA during that expiration (CSA: breath-3, 79 +/- 3% to 62 +/- 6%; breath-2, 76 +/- 4% to 50 +/- 10%; breath-1, 64 +/- 8% to 36 +/- 10%). Occlusion occurred at a pressure significantly less than that generated during the previous unoccluded breath (Pes: breath-1, -10.8 +/- 2.9 cm H2O; occlusion, -8.2 +/- 1.9 cm H2O). These results show that expiratory narrowing produced a significant reduction of CSA at end-expiration prior to obstructive apnea.