Nasally administered midazolam appears to be a useful method for rapidly sedating children prior to the induction of anesthesia. We determined the peak plasma concentrations after intranasal administration of midazolam and compared this to plasma concentrations achieved after intravenously administered midazolam in 18 children between the ages of 14 months and 5 yr, who underwent elective closure of an asymptomatic atrial septal or ventricular septal defect. Preanesthetic medication was at the discretion of the attending anesthesiologist. Induction of anesthesia was with halothane in N2O and O2 via mask, and tracheal intubation was performed after the administration of fentanyl or sufentanil plus pancuronium. Anesthesia was maintained with these agents, and augmented with halothane or isoflurane. As soon as arterial access was established, the patient received 0.1 mg/kg of either intranasal or intravenous midazolam. Midazolam concentrations were measured by gas chromatography-mass spectrometry. Intranasal midazolam achieved its peak plasma concentration of 72.2 +/- 27.3 ng/ml in 10.2 +/- 2 min. Ten minutes after the administration of midazolam, the mean plasma concentration in the intranasal midazolam group was 57% of the concentrations in the group receiving midazolam intravenously. These results confirm the clinical impression that intranasal administration of midazolam rapidly achieves sedative plasma concentrations in children.