A randomised, single-blinded study was conducted to compare patient-controlled epidural analgesia with continuous infusion epidural analgesia for the treatment of pain following post-traumatic pelvic reconstruction. The patient-controlled group (n = 11) received a background infusion of 4 ml.h-1 of bupivacaine 0.125% with fentanyl 1 microgram.ml-1, and 3-6 ml bolus doses, self administered, as required (with a 15 min lockout interval). The continuous infusion group (n = 12) received a continuous infusion of the same solution through an identical apparatus, but with the demand button deactivated. This was started at 10 ml.h-1 and adjusted by the anaesthetist, as required, up to a maximum of 25 ml.h-1. Pain scores, side effects, and the volumes of drug infused were recorded over the first 3 postoperative days. One patient from each group was withdrawn because of catheter-related problems. Pain scores were similar and the incidence of nausea and pruritus was low in both groups. There was no recorded instance of respiratory depression or hypotension and there was no significant difference between the groups in the volumes of drug solution received. Patient satisfaction was equally very good in both groups. Patient-controlled epidural analgesia is an effective means of providing pain relief after post-traumatic pelvic reconstruction, but did not significantly reduce analgesic requirements in comparison with continuous infusion epidural analgesia.