We performed a double-blind study of the dose-response relationship of intrathecal morphine (0, 0.3, 1, and 2.5 mg) for postoperative pain relief in 33 subjects who underwent total knee or hip replacement surgery. Assessments commenced 1 hour after the opioid injection, which was given at the end of surgery, and continued for 24 hours. Pain measurements, supplementary analgesia requirements, and adverse effects were recorded. Intrathecal morphine provided effective, long-lasting pain relief. All doses delayed the initial perception of discomfort (T-Pain) and also postponed the onset of severe pain requiring analgetic supplementation (T-Morphine) (1.25 hours control with placebo injections; greater than 20 hours with intrathecal morphine 0.3, 1, and 2.5 mg: P less than 0.05). Although 0.3 mg usually provided good analgesia it was unsatisfactory in three of 10 patients (30%), whereas 1 and 2.5 mg were absolutely reliable. Respiratory depression (increased PaCO2), common after the administration of 1 or 2.5 mg intrathecal morphine, was slow in onset and prolonged. The respiratory depression after 2.5 mg was more profound than after 1 mg, and produced apnea necessitating large-dose naloxone therapy. Pruritus was unique to intrathecal morphine administration, but nausea, vomiting, and urinary retention were common in all the groups. We conclude that no ideal dose of intrathecal morphine exists because, even with small quantities, minor adverse effects are evident. Doses between 0.3 and 1 mg, however, should provide good analgesia free from the major complication, respiratory depression.