The purpose of this randomized double-blind study was to determine the optimal dose of epidural morphine by establishing a dose-effect relationship. The 139 patients, who had orthopedic operations on the lower extremities, received continuous lumbar epidural anesthesia with bupivacaine, 0.75%, with or without the addition of 1, 2, 3, 4, or 5 mg of morphine hydrochloride. Analgesia and side effects were determined during the first 24 hr postoperatively. In the 12-hr period after epidural anesthesia, arterial blood gas tensions were compared between those patients who received 5 mg morphine (n = 13) and those who received no morphine (n = 14). Patients who received 2 or more mg of morphine were less likely to require the administration of postoperative systemic analgesics (P less than 0.05). The addition of 2 or more mg of morphine to bupivacaine, 0.75%, reduced postoperative pain intensity (P less than 0.05); 5 mg of morphine reduced pain intensity for the longest time. Frequency of catheterization and pruritus increased dose-dependently. The mean PaCO2 after 5 mg of epidural morphine averaged 5 mm Hg higher than in the control group, indicating minor respiratory depression, better analgesia, or both. The dose of 3 mg of epidural morphine added to the local anesthetic is recommended for postoperative analgesia after surgery of the lower extremity; it is a compromise that provides adequate analgesia with an acceptably low frequency and intensity of side effects.