The postoperative analgesic effects of intra-articular injections of bupivacaine and/or morphine were examined prospectively in 437 patients who had total knee replacement for osteoarthritis. They were divided randomly into four groups. Group I received 10 mg of morphine (1 ml) and 9 ml of saline, group II received 10 ml of bupivacaine (2.5 mg/ml), group III received 10 ml of saline, and group IV received 10 mg of morphine (1 ml) and 9 ml of bupivacaine (2.5 mg/ml). All analgesics administered in the first 24 hours after operation were recorded. The patients rated their pain on the McGill-Melzack scale at 1, 6, 12 and 24 hours. No significant differences were found between any of the groups in the use of Demoral and/or Toradol in 24 hours, the length of stay in hospital or the pain rating at 1, 6, 12 or 24 hours. Patients in groups I and IV, whose injections included morphine, used significantly more morphine in the first 24 postoperative hours than did groups II or III.