Eighty patients premedicated with temazepam were allocated randomly to receive an intravenous bolus of midazolam (either 0, 1, 2 or 4 mg) 4 min before a propofol infusion using a system designed to maintain a target blood concentration of 3 micrograms.ml-1. Anaesthesia was successfully induced within 3 min of starting the infusion in 45% of patients receiving no midazolam. This rose to 70%, 85% and 95% in the groups receiving 1, 2 and 4 mg midazolam boluses respectively (p < 0.005). There were reductions in arterial blood pressure 3 min after starting the infusion in all groups but there was no significant difference between the groups. Apnoea was rare. Small doses of midazolam usefully enhance the effect of a propofol infusion administered at a dose which by itself would not induce anaesthesia in a majority of patients.