The respiratory effects of 2 postoperative analgesic regimens were compared in 2 groups of 16 patients each, recovering from general anesthesia and major surgery. One group received a pain-relieving dose of i.v. morphine (mean 18.1 mg), with the same dose repeated as a continuous i.v. infusion over the subsequent 24 h. The other group received regional anesthesia using bupivacaine. The patients were monitored for 16 h after surgery. The 2 analgesic regimens provided patients with comparable analgesia throughout the study period, but there were quite different respiratory effects in the 2 groups. Ten patients receiving morphine infusions had a total of 456 episodes of pronounced O2 desaturation (SaO2 < 80%). These occurred only while the patients were asleep, and all were associated with disturbances in ventilatory pattern, namely, obstructive apnea (144 episodes in 8 patients), paradoxic breathing (275 episodes in 6 patients) and period of slow ventilatory rate (37 episodes in 1 patient). In contrast, in patients receiving regional anesthesia, O2 saturation never decreased below 87%. Central apnea, obstructive apnea and paradoxic breathing occurred more frequently in patients in the morphine group (12, 10 and 10 patients, respectively) than patients in the regional anesthesia group (4, 3 and 5 patients, respectively). The interaction of sleep and morphine analgesia produced disturbances in ventilatory pattern, causing profound O2 destruction. Postoperative pain relief using regional anaesthesia has a greater margin of safety in terms of respiratory side effects than does the continuous administration of opiates.