We undertook a prospective study to evaluate the relationship between the onset and degree of sedation and the midazolam plasma concentration in children between 6 and 18 years of age during esophagogastroduodenoscopy. Thirteen boys and seven girls (median age 13.5 years) were studied. Midazolam was injected intravenously for 5 minutes, and the dose was titrated to sedation or a maximum dose of 0.1 mg/kg was given. Plasma midazolam concentration was determined just before and at 5, 10, 15, 30, 45, and 60 minutes after the start of midazolam injection. The patient's level of sedation was evaluated by an assistant at each blood sampling time. Clearance, volume of distribution, and terminal elimination (beta) half-life were estimated from a biexponential fit of the serial plasma midazolam concentrations. Mean beta half-life of midazolam was 47 +/- 26 minutes and mean clearance was 10.0 +/- 5.0 ml/min per kilogram of body weight. Maximum level of sedation occurred at 5 minutes after initiation of the injection and corresponded to a mean peak midazolam serum concentration of 229 +/- 39 micrograms/L. Thereafter, a decline of mean sedation scores paralleled the decrease in midazolam concentration. Mean oxygen saturation remained greater than 94% during the study. We conclude that children metabolize and excrete midazolam more rapidly than adults do and that sedation adequate for endoscopy is safely achieved in the majority of children with a midazolam dose of 0.05 to 0.1 mg/kg and a mean peak midazolam concentration greater than 200 microgram/L.