The use of lidocaine in spinal anaesthesia is associated with transient neurological syndrome (TNS). Bupivacaine has a lower incidence of TNS as an alternative but it may have a prolonged action. This study systematically reviews the literature about the recovery profile of patients undergoing spinal anaesthesia, using bupivacaine for arthroscopic knee surgery. We identified 17 eligible randomized clinical trials (RCTs) (1268 patients). All the articles in this review, except one, used hyperbaric bupivacaine. Five trials compared different doses of bupivacaine (range 3-15 mg). Large doses of bupivacaine (10 and 15 mg) were associated with delayed recovery, and supine positioning was associated with a high incidence of failure. With unilateral positioning, a dose as low as 4-5 mg seems to be sufficient. Five trials comparing bupivacaine or levobupivacaine with ropivacaine showed no significant difference in the time to home discharge. When bupivacaine was combined with fentanyl in two trials, marginal delay in recovery was found [time to discharge (min); weighted mean difference (WMD) 14.1, 95% CI 11.9-40.1] and increased nausea and pruritus but had reduced postoperative pain. Unilateral and bilateral spinal anaesthesia were assessed in two trials, and the latter group was associated with early recovery and discharge [time to discharge (min); WMD -41.6, 95% CI -63.6 to -19.6). The results of our systematic review suggest that 4-5 mg of hyperbaric bupivacaine can effectively produce spinal anaesthesia for knee arthroscopy with unilateral positioning. Ropivacaine or the addition of adjuvants did not improve the recovery time. There is a need for tighter RCTs with more consistent endpoints.