Opioids are widely used as additives to local anesthetics for intrathecal anesthesia. Benefit and risk remain unclear. We systematically searched databases and bibliographies to February 2011 for full reports of randomized comparisons of any opioid added to any intrathecal local anesthetic with the local anesthetic alone in adults undergoing surgery (except cesarean section) and receiving single-shot intrathecal anesthesia without general anesthesia. We included 65 trials (3338 patients, 1932 of whom received opioids) published between 1983 and 2010. Morphine (0.05-2mg) and fentanyl (10-50 μg) added to bupivacaine were the most frequently tested. Duration of postoperative analgesia was prolonged with morphine (weighted mean difference 503 min; 95% confidence interval [CI] 315 to 641) and fentanyl (weighted mean difference 114 min; 95% CI 60 to 168). Morphine decreased the number of patients needing opioid analgesia after surgery and decreased pain intensity to the 12th postoperative hour. Morphine increased the risk of nausea (number needed to harm [NNH] 9.9), vomiting (NNH 10), urinary retention (NNH 6.5), and pruritus (NNH 4.4). Fentanyl increased the risk of pruritus (NNH 3.3). With morphine 0.05 to 0.5mg, the NNH for respiratory depression varied between 38 and 59 depending on the definition of respiratory depression chosen. With fentanyl 10 to 40 μg, the risk of respiratory depression was not significantly increased. For none of these effects, beneficial or harmful, was there evidence of dose-responsiveness. Consequently, minimal effective doses of intrathecal morphine and fentanyl should be sought. For intrathecal buprenorphine, diamorphine, hydromorphone, meperidine, methadone, pentazocine, sufentanil, and tramadol, there were not enough data to allow for meaningful conclusions.