OBJECTIVE A benzodiazepine is generally used when sedation is required for endoscopy, whilst the newer agent propofol appears to have a more suitable pharmacokinetic profile. The aim of this study was to compare the quality of sedation provided by midazolam and propofol under controlled conditions. METHODS Ninety healthy outpatients requesting sedation at diagnostic gastroscopy were prospectively randomized to receive a bolus dose of either midazolam 0.06 mg/kg or propofol 0.6 mg/kg, followed by repeat doses of 50% of the initial dose as required. No topical anesthesia was used. The endoscopist and the patient replied to a questionnaire. RESULTS Patients receiving propofol tolerated endoscopy better (p = 0.01); reached a deeper maximum level of sedation (p = 0.001); and recovered more rapidly (p = 0.001). There was a similar frequency of amnesia for the procedure (propofol 67% vs. midazolam 65%); of oxygen desaturation (four patients in each group, < 85%); and of perceived patient comfort (p = 0.5). CONCLUSIONS Compared to midazolam, propofol facilitated gastroscopy to a greater extent. However, due to its narrower therapeutic range, propofol is the more demanding agent to administer, thus making it less universally applicable than midazolam.