In an effort to establish the minimum effective dose of theophylline in the treatment of idiopathic apnea of prematurity, a prospective trial of 22 infants with at least 0.33 episodes of apnea per hour were studied. Apnea was diagnosed exclusively by continuous recording of heart rate, respiratory impedance, end-tidal CO2, and either or both transcutaneous oxygen and pulse oximetry. Four discrete serum concentrations of theophylline (23 mumol/l or 4.2 mg/L, 47 mumol/L or 8.5 mg/L, 70 mumol/L or 12.7 mg/L, and 84 mumol/L or 15.3 mg/L) were attained by using repeated loading doses of 4 mg/kg and increasing the maintenance dose from 1 to 1.5 mg/kg to 2 to 2.5 mg/kg, given every 8 hours. Before treatment and 24 hours after each loading dose, airway occlusions and measures of tidal volume, minute ventilation, and respiratory timing were performed. The effectiveness of therapy was assessed by either a continuous computer data-acquisition system or paper recording for the duration of the study. Of the 22 infants, three responded at level 1, three at level 2, and 10 at level 3. One of the four infants loaded to the fourth level had a sustained response for a total cumulative response of 77%. The five remaining infants required additional treatment with doxapram or continuous positive airway pressure. There was a significant increase in inspiratory pressure 100 msec after airway occlusion, maximum inspiratory pressure during airway occlusion, tidal volume, ratio of tidal volume to inspiratory time (mean inspiratory flow), and minute ventilation from the pretreatment measurements to those at the maximum dose of theophylline. The apnea response did not correlate with these improvements in ventilation measures.