Although obstructive sleep apnea (OSA) occurs in the pediatric population, diagnostic criteria have not been established. Since criteria for adult OSA are well established, we asked whether commonly used adult criteria, such as the apnea index (based on the number of obstructive apnea [OA] events per hour), would identify children with serious sleep-related upper airway obstruction. Polysomnographic data were analyzed from 20 children (ages 8 months to 16 yr) with clinical evidence of upper airway obstruction during sleep (loud snoring and labored breathing) and who had cyclic oscillations of oxyhemoglobin saturation (SaO2) during sleep. The overnight studies included sleep state (EEG, EOG, and EMG), SaO2, ECG, nasal (end-tidal CO2) and oral (thermistor) airflow, chest and abdominal movement (inductance plethysmography), and video camera and behavioral observations. Measurements included the number of obstructive events > or = 10 s, the number of desaturations (> or = 5% decrease lasting > or = 5 s), the number of desaturation episodes to < 90%, < 85%, and < 80% lasting > 5 s, and the percentage of sleep time with SaO2 values < 90%. Gas exchange was impaired as evidenced by cyclic decreases in SaO2 and elevated PETCO2 values (maximum value 58 +/- 6 mm Hg). The children experienced 175 +/- 168 (range 6 to 609) episodes of decreased SaO2 > 5%, with an average minimum SaO2 of 66 +/- 13% (range 30 to 85%). The average number of apnea events was only 1.9 +/- 3.2 events/h (range 0 to 10.4).(ABSTRACT TRUNCATED AT 250 WORDS)