Upper gastrointestinal endoscopy is routinely used to follow up patients with digestive disease. The influence of premedication with diazepam on the procedure and on patients' acceptance of a second endoscopy was evaluated in a double-blind trial conducted on 81 patients randomized to 3 types of intravenous premedication: placebo (group A, 29 patients), diazepam 5 mg (group B, 27 patients) and diazepam 10 mg (group C, 25 patients). Endoscopy was performed under unfavourable conditions (refusal or agitation) in 12, 4 and 1 patients respectively of groups A, B and C (P less than 0.02). Out of 65 patients questioned one month later, 1 (group A) sternly refused a second endoscopy, 10 accepted unreservedly (O in group A, 4 in group B, 6 in group C; P less than 0.05 between A and C) and 54 accepted reluctantly. To find out whether diazepam was beneficial by impairing memory, a memory score on duration of the procedure was established; there were no significant differences between the three groups (P less than 0.001), but they did not correlate with the patients' acceptance of a second endoscopy. It is concluded that premedication with diazepam 10 mg i.v. makes upper gastrointestinal endoscopy easier to perform and facilitates acceptance of a repeat endoscopy 1 month later. This effect is probably independent of the impairment in memory induced by diazepam.