Nasal administration of sufentanil or midazolam is effective for preinduction of pediatric patients, but there are no data on which to base a choice between them. This blinded randomized study compares behavioral and physiologic responses to sedation with one of these medications followed by inhalation or intravenous induction. Ninety-five patients aged 0.5-10 yr scheduled for elective surgery were stratified by age: 30 infants 0.5-2 yr, 38 preschoolers 2.1-5 yr, and 27 school-age children 5.1-10 yr. They were randomized to receive 0.04 ml/kg of midazolam (0.2 mg/kg) or sufentanil (2 micrograms/kg). Hemoglobin oxygen saturation by pulse oximetry (SpO2) and sedation score were recorded prior to drug administration, at 2.5-min intervals for 10 min, at separation, and during induction with graded halothane in oxygen. Intubation was performed under deep halothane or 3 mg/kg of thiopental and 0.1 mg/kg of pancuronium. Chest wall compliance was assessed qualitatively in all patients prior to intubation. To assess the effects of a mild standardized stress on unpremedicated patients, 75 of the children with parents present were scored before and after oximeter probe placement: of these, in 63% the sedation score did not change; 33% appeared more anxious; and only 4% seemed reassured. Children of all ages reacted negatively to physicians, and 23% were crying prior to administration of drugs. Sufentanil appeared less unpleasant to receive than midazolam: children cried 46 +/- 100 versus 76 +/- 73 s (P less than 0.05), respectively, but by 7.5 min, no child was crying. Median behavior scores at maximum anxiolysis were not different, but response to sufentanil was more variable.(ABSTRACT TRUNCATED AT 250 WORDS)