We studied 14 subjects who were selected to represent the broad range of severity of snoring: group A, four subjects who gave a history of snoring only when provoked by nasal obstruction or alcohol intake; group B, six subjects who typically snored for long periods each night; and group C, four subjects who snored heavily all night and who typically experienced a few episodes of obstructive apnea (mean apnea index 4 apneas/h). Low levels of nasal continuous positive airway pressure (CPAP) (range, 2.0-6.0 cmH2O; mean, 4.0 cmH2O) prevented snoring. Nasal occlusion caused upper airway closure during inspiratory efforts in all 14 subjects. There was a relationship between the clinical severity of snoring and the upper airway closing pressure (UACP). Upper airway closure occurred at greater suction pressures in group A than in group C but there was overlap between the three categories. The upper airway was consistently more collapsible in rapid-eye-movement sleep than in non-rapid-eye-movement sleep. There was little evidence of breath-by-breath improvement of upper airway stability during sustained asphyxia, the UACP remaining constant despite marked increases in drive to the diaphragm. In five subjects UACP was measured following alcohol intake. Alcohol reduced upper airway stability in all subjects in a dose-dependent manner.