This prospective controlled study was aimed at evaluating the efficacy of caffeine in treating recurrent idiopathic apnea in the premature infant. Eighteen preterm infants (29 to 35 weeks' gestation) were studied. Recordings during the first 24 hours and on the fifth day of caffeine treatment showed a significant decrease of severe apnea (P less than 0.01) and of mild apnea (P less than 0.001) in the treated group (group 1) as compared with the control group (group II). No treatment of apnea other than caffeine was required in group I, whereas six neonates in group II had such severe and frequent apneic episodes for more than 48 hours that withholding additional treatment was believed to be unethical. No undesirable side effects of caffeine treatment were observed.