We performed a prospective, randomized, double-blind study to evaluate the efficacy of the currently recommended low doses of midazolam for conscious sedation compared with diazepam for colonoscopy. Each agent was administered in a fixed ratio dose in combination with meperidine, and titrated incrementally to allow for adequate sedation prior to initiating and during the procedure. The currently recommended starting dose of midazolam (0.03 mg/kg) proved to be very appropriate for pre-medication. In contrast, the currently recommended starting dose of diazepam (0.10 mg/kg) proved excessive in 21% of patients (especially in those aged > 65). The low initial and incremental doses of midazolam compared favorably with diazepam in all efficacy parameters studied and exceeded diazepam in post-procedure amnesia scores (p = 0.01). Moreover, the sedative effects of midazolam at these lower doses were not lost despite long duration procedures (> 40 min). We conclude that midazolam, given in small incremental doses, in combination with meperidine, produces effective conscious sedation for colonoscopy and exceeds diazepam in its amnestic effect.