Sixty children undergoing inguinal or urogenital surgery were allocated randomly to three groups to receive a caudal injection of either 0.125% bupivacaine 0.75 mL kg-1 with 0.5% midazolam 50 micrograms kg-1 (n = 20) or with 1% morphine chlorhydrate 0.05 mg kg-1 (n = 20), or bupivacaine alone (n = 20) after surgery under general anaesthesia. There were no significant changes in heart rate, blood pressure, respiratory rate or oxygen haemoglobin saturation values in all groups, and there were no significant differences in the incidence of vomiting and pruritus between the groups (P > 0.05). Sedation scores were higher in the bupivacaine-midazolam and the bupivacaine-morphine groups than in the bupivacaine group at 8-12 h post-operatively (P < 0.01). The durations of analgesia were 21.15 +/- 1.2 h in the bupivacaine-midazolam group, 14.50 +/- 1.6 h in the bupivacaine-morphine group and 8.15 +/- 1.3 h in the bupivacaine group. Differences between the bupivacaine-midazolam group and the bupivacaine group (P < 0.001), the bupivacaine-midazolam group and the bupivacaine-morphine group (P < 0.01), and the bupivacaine-morphine group and the bupivacaine group (P < 0.01) were significant. It is suggested that caudal administration of a bupivacaine-midazolam mixture produces a longer duration of post-operative analgesia than a bupivacaine-morphine mixture and bupivacaine alone with sedation for 8-12 h post-operatively.