Apnea in preterm infants has been classified as obstructive, central (nonobstructive), and mixed, based on the presence or absence of upper airway obstruction. Continuous positive airway pressure (CPAP) is widely used in apneic infants, although its mechanism of action is still unclear. To determine whether CPAP is equally effective in obstructive and nonobstructive apnea, we compared the types of apnea observed in 14 preterm infants during sequential 45-minute periods with and without CPAP. CPAP markedly decreased the incidence of both mixed and obstructive apnea episodes of greater than or equal to 5 seconds (P less than 0.01 and less than 0.03, respectively). In contrast, central apnea episodes of greater than or equal to 5 seconds were entirely unaffected by CPAP. Although minute ventilation was unchanged, transcutaneous PO2 increased by 11 +/- 11 mm Hg during CPAP whether or not apnea was present. We postulate that CPAP reduces apnea in preterm infants by relief of upper airway obstruction, possibly via splinting of the pharyngeal airway.