Efficacies of three alternate methods of postoperative analgesia were studied in 156 patients who had total knee arthroplasty (TKA). Forty-two of these patients received parenteral meperidine hydrochloride or morphine (Group 1), 58 patients received periodic epidural injections of morphine (Group 2), and 56 patients received continuous epidural infusions of bupivacaine hydrochloride and Duramorph (Group 3). The postoperative course of all patients was documented in terms of the incidence and severity of pain, range of joint motion, duration of hospitalization, and occurrence of complications. Although epidural analgesia increased the cost and duration of the operation, good-to-excellent pain relief was attained in 86% (Group 2) and 88% (Group 3) of cases with epidural analgesia compared with 61% of patients (Group 1) receiving conventional analgesia. Moreover, 67% of patients in Group 1 experienced frequent episodes of moderate-to-severe postoperative pain in contrast to 40% of patients in Group 2 and only 10% of patients in Group 3. As a result of diminished pain, greater joint motion was obtained within the first 72 hours in Groups 2 and 3. They also had shorter hospitalization (9.6 days versus 11.2 days for Group 1 and 10.8 days for Group 2). However, the use of epidural analgesia did not reduce the incidence of complications, including nausea. Continuous infusion of epidural bupivacaine and Duramorph provided good-to-excellent control of postoperative pain after TKA. However, better analgesics are needed to reduce the high incidence of side effects associated with various treatment methods.