We reviewed systematically the published evidence on the effectiveness of antibiotic prophylaxis for the reduction of wound infection in patients undergoing total hip and total knee replacement. Publications were identified using the Cochrane Library, MEDLINE, EMBASE and CINAHL databases. We also contacted authors to identify unpublished trials. We included randomised controlled trials which compared any prophylaxis with none, the administration of systemic antibiotics with that of those in cement, cephalosporins with glycopeptides, cephalosporins with penicillin-derivatives, and second-generation with first-generation cephalosporins. A total of 26 studies (11 343 participants) met the inclusion criteria. Methodological quality was variable. In a meta-analysis of seven studies (3065 participants) antibiotic prophylaxis reduced the absolute risk of wound infection by 8% and the relative risk by 81% compared with no prophylaxis (p < 0.00001). No other comparison showed a significant difference in clinical effect. Antibiotic prophylaxis should be routine in joint replacement but the choice of agent should be made on the basis of cost and local availability.