The effectiveness of intra-articular (IA) morphine in arthroscopic procedures of the knee joint was analysed in all randomised and controlled trials that included injections of morphine and placebo into the knee joint, and where the data were analysable. Sensitivity of the studies and effectiveness were analysed for three different periods: immediate (0-2h), early (2-6h) and late (6-30h). Sensitivity for each period was assumed if pain intensity was at least 30% of the maximum of 100 on the visual analogue scale in the placebo group. Six different doses (1-10mg) of IA morphine were compared with placebo. The injections were made at the end of surgery, before the arthroscope was removed from the joint. In the immediate period 7/15 sensitive trials were positive, in the early period 8/12 sensitive trials were positive and in the late period 10/13 sensitive trials were positive. Most positive studies had used higher doses (3-5mg) compared with negative studies that had mainly used 1mg. Two studies using patient controlled analgesia consumption of analgesics as an outcome were also positive. The only sensitive study of four dose-response comparisons indicated that 5mg of IA morphine was more effective than 1mg. The only sensitive study of three cross-route comparisons showed no difference between 5mg of IA and 5mg of intra-muscular morphine. All insensitive trials, including placebo (except two individual comparisons), cross-route and dose-response comparisons, were negative. The analysis of sensitive studies indicates that 5mg of IA morphine injected into the knee joint provides postoperative pain relief for up to 24h. A minimum of 30% of the maximum possible pain intensity is needed for an analgesic effect to be detected in a study.