To determine the induction and maintenance characteristics of desflurane in pediatric patients, the authors anesthetized 206 infants and children aged 1 month to 12 yr with nitrous oxide plus desflurane and/or halothane in oxygen. Patients were assigned to one of four groups: anesthesia was 1) induced and maintained with desflurane after premedication with an oral combination of meperidine, diazepam, and atropine; 2) induced and maintained with desflurane; 3) induced with halothane and maintained with desflurane; or 4) induced and maintained with halothane. An unblinded observer recorded time to loss of consciousness (lid reflex), time to intubation, and clinical characteristics of the induction and maintenance of anesthesia. Moderate-to-severe laryngospasm (49%) and moderate-to-severe coughing (58%) occurred frequently during induction of anesthesia with desflurane; the incidence of these was not altered by premedication. In contrast, laryngospasm and coughing were rare during induction of anesthesia with halothane. In unpremedicated patients, time to loss of lid reflex (mean +/- SD) was similar for desflurane (2.4 +/- 1.2 min) and halothane (2.1 +/- 0.8 min). During induction of anesthesia, before laryngoscopy and intubation, mean arterial pressure less than 80% of baseline was more common with halothane; heart rate and mean arterial pressure greater than 120% of baseline were more common with desflurane. Intraoperatively, heart rate greater than 120% of baseline was more common with desflurane; blood pressures were similar for the two anesthetics. The authors conclude that the high incidence of airway complications during induction of anesthesia with desflurane limits its utility for inhalation induction in pediatric patients. Anesthesia can be safely maintained with desflurane if induced with a different anesthetic.