Twenty-two patients were studied while receiving epidural analgesia with diamorphine after major lower abdominal surgery under combined regional and general anaesthesia. Epidural PCA began when the intra-operative epidural block with bupivacaine wore off enough for the patient to request treatment. It was started with 2 mg of diamorphine and continued with a reducible background infusion that was initially set at 0.2 mg h−1 and supplemented by on-demand doses of 0.2 mg, with a lockout time of 15 min. The patients received routine post-operative monitoring and care, with pain at rest being assessed on a four-point verbal rating scale (VRS, none, mild, moderate, severe) at 5, 10, 15, 30, 45, 60, 90 and 120 min from the start of ePCA, then hourly until 24 h and then 2-hourly until 48 h. VRS on coughing and a 10 cm visual analogue score (VAS) at rest and on coughing were recorded at the same times at 4 h, then 4 hourly until 24 h and then at 48 h, at which times, blood samples were also taken to measure morphine concentrations by radioimmunoassay. Analgesia started promptly and reached a maximum at between 30 and 45 min, accompanied by maximum sedation. Thereafter clinically acceptable analgesia was maintained without undue sedation for 48 h, though pain on coughing was less well controlled than pain at rest. After the initial loading dose of diamorphine, the 95%confidence intervals (CI) for further consumption were 3.7 to 17 mg (average 9.7) in the first 24 h and 2.1 to 12.9 mg (average 6.7 mg) in the second 24 h. The plasma morphine concentrations rose to a plateau by about 15 min, with concentrations within 95% CI from 0 to 11 ng mL−1 (average 5 ng mL−1). The VRS and VAS pain scores were analysed by a conservative approach that treated them as ordinal data, and by a parametric approach that treated them as interval data. Both approaches conveyed broadly similar information about the post-operative analgesia.