Continuous passive motion after major knee surgery optimizes the functional prognosis but causes severe pain. The authors tested the hypothesis that postoperative analgesic techniques influence surgical outcome and the duration of convalescence. Before standardized general anesthesia, 56 adult scheduled for major knee surgery were randomly assigned to one of three groups, each to receive a different postoperative analgesic technique for 72 h: continuous epidural infusion, continuous femoral block, or intravenous patient-controlled morphine (dose, 1 mg; lockout interval, 7 min; maximum dose, 30 mg/4 h). The first two techniques were performed using a solution of 1% lidocaine, 0.03 mg/ml morphine, and 2 microg/ml clonidine administered at 0.1 ml x kg(-1) x h(-1). Pain was assessed at rest and during continuous passive motion using a visual analog scale. The early postoperative maximal amplitude of knee flexion was measured during continuous passive motion at 24 h and 48 h and compared with the target levels prescribed by the surgeon. To evaluate functional outcome, the maximal amplitudes were measured again on postoperative day 5, at hospital discharge (day 7), and at 1- and 3-month follow-up examinations. When the patients left the surgical ward, they were admitted to a rehabilitation center, where their length of stay depended on prospectively determined discharge criteria The continuous epidural infusion and continuous femoral block groups showed significantly lower visual analog scale scores at rest and during continuous passive motion compared with the patient-controlled morphine group. The early postoperative knee mobilization levels in both continuous epidural infusion and continuous femoral block groups were significantly closer to the target levels prescribed by the surgeon than in the patient-controlled morphine group. On postoperative day 7, these values were 90 degrees (60-100 degrees)(median and 25th-75th percentiles) in the continuous epidural infusion group, 90 degrees (60-100 degrees) in the continuous femoral block group, and 80 degrees (60-100 degrees) in the patient-controlled morphine group (P < 0.05). The durations of stay in the rehabilitation center were significantly shorter: 37 days (range, 30-45 days) in the continuous epidural infusion group, 40 days (range, 31-60 days) in the continuous femoral block group, and 50 days (range, 30-80 days) in the patient-controlled morphine group (P < 0.05). Side effects were encountered more frequently in the continuous epidural infusion group. Regional analgesic techniques improve early rehabilitation after major knee surgery by effectively controlling pain during continuous passive motion, thereby hastening convalescence.