Intrathecal morphine without local anaesthetic is often added to a general anaesthetic to prevent pain after major surgery. Quantification of benefit and harm and assessment of dose-response are needed. We performed a meta-analysis of randomized trials testing intrathecal morphine alone (without local anaesthetic) in adults undergoing major surgery under general anaesthesia. Twenty-seven studies (15 cardiac-thoracic, nine abdominal, and three spine surgery) were included; 645 patients received intrathecal morphine (dose-range, 100-4000 microg). Pain intensity at rest was decreased by 2 cm on the 10 cm visual analogue scale up to 4 h after operation and by about 1 cm at 12 and 24 h. Pain intensity on movement was decreased by 2 cm at 12 and 24 h. Opioid requirement was decreased intraoperatively, and up to 48 h after operation. Morphine-sparing at 24 h was significantly greater after abdominal surgery {weighted mean difference, -24.2 mg [95% confidence interval (CI) -29.5 to -19.0]}, compared with cardiac-thoracic surgery [-9.7 mg (95% CI -17.6 to -1.80)]. The incidence of respiratory depression was increased with intrathecal morphine [odds ratio (OR) 7.86 (95% CI 1.54-40.3)], as was the incidence of pruritus [OR 3.85 (95% CI 2.40-6.15)]. There was no evidence of linear dose-responsiveness for any of the beneficial or harmful outcomes. In conclusion, intrathecal morphine decreases pain intensity at rest and on movement up to 24 h after major surgery. Morphine-sparing is more pronounced after abdominal than after cardiac-thoracic surgery. Respiratory depression remains a major safety concern.