Sixty patients, scheduled for Caesarean section were randomly allocated to receive by the epidural route in a double-blind fashion one of the following patient-controlled analgesia mixtures for the relief of postoperative pain: sufentanil 2 micrograms.ml-1 in 0.9% sodium chloride, sufentanil 2 micrograms.ml-1 + adrenaline 2.5 micrograms.ml-1, or sufentanil 2 micrograms.ml-1 + clonidine 3 micrograms.ml-1. Patient-controlled analgesia settings were a basal infusion rate of 2.5 ml.h-1, an incremental dose of 2.5 ml, a lockout interval of 10 min and a 1-h limit of 10 ml. Whereas patient demographics and pain scores between the groups were not different, the 24-h consumption of sufentanil was significantly lower in the groups receiving a combination (167.5 SD 45 and 139.1 SD 31.9 micrograms for the adrenaline and clonidine groups respectively) as compared to the plain sufentanil regimen (208.2 SD 38.9 micrograms). Although sufentanil requirements were the lowest in the clonidine admixture group, there were no differences with regard to sedation as compared to the plain sufentanil group. The quality of sleep appeared to be significantly better in the sufentanil/adrenaline group despite a significantly lower degree of sedation and higher incidence of pruritus. Treatment of pruritus with naloxone did not seem to influence the quality of analgesia.