We performed a prospective, randomized, double-blind study of 41 children (6-18 years of age) who were undergoing esophagogastroduodenoscopy (EGD) to compare the efficacy of diazepam and midazolam with respect to quality of sedation and amnesia. The endoscopist assessed the patients for control of salivation, gag reflex, vomiting and cooperation after intravenous injection of either 0.1-0.15 mg/kg of midazolam or 0.2-0.4 mg/kg of diazepam. The patients answered a questionnaire at 1 and 24 h after the procedure to assess recall of procedure details, pain and/or discomfort, and their medication choice for future procedures. Vital signs were monitored for 1 h after injection. Although midazolam caused greater mean maximum increase in heart rate than diazepam (30 vs. 14/min), no difference was found between the two treatment groups with respect to other vital signs. There was also no difference in physician's assessment or patient recall of specific events during the procedure. However, significantly fewer patients recalled pain or discomfort with midazolam at both 1 and 24 h following the procedure (p = 0.02). In addition, more patients receiving midazolam indicated preference for the same sedation for future procedures. We conclude that midazolam may provide better amnesia in children undergoing endoscopic procedures.