Aim:A systematic review of controlled trials of therapy of Clostridium difficile intestinal infection using methodology described by the Cochrane Collaboration.Methods:Trials were identified by searching computer databases over the years 1978–1996. Trials were included if they were (a) prospective randomized, controlled trials and (b) included patients with symptomatic disease. The primary end‐point was clinical resolution of diarrhoea. Secondary end‐points were clinical relapse and stool clearance of C. difficile and C. difficile toxin.Results:Nine trials (469 patients) satisfying the inclusion criteria were identified. Two trials were placebo controlled. Six trials compared vancomycin to other antibiotics (fusidic acid, bacitracin, teicoplanin and metronidazole). For clinical resolution response rates ranged from 21 (placebo) to 100% (vancomycin). On pooling the trials, no antibiotic showed clear therapeutic superiority. Rates of clinical relapse ranged from 5 to 42%. Only one trial showed significant advantage of one antibiotic over another for prevention of relapse (teicoplanin vs. fusidic acid).Conclusion:The published data are limited, and further studies are required.