One hundred patients for endoscopy, aged, between 18 and 74 years were randomly allocated to two equal groups. Group A received 10 mg diazepam intravenously rapidly over 2-4 s; Group B received diazepam intravenously over 1-2 min, titrated to provide a satisfactory level of sedation (mean dose 15.9 mg). Patient co-operation during endoscopy was similar in both groups, but one patient in Group A developed respiratory depression. Tests of psychomotor function after endoscopy showed greater impairment 30 min following injection for Group B, compared with Group A (P less than 0.02). Total or partial amnesia for endoscopy was present in 62% of Group A, and 90% in Group B (P less than 0.005). However, postal follow-up showed that endoscopy was acceptable to 96% of Group A and 98% of Group B. Delayed sedation later in the day was reported by 26% of Group A compared with 48% of Group B.