One hundred and seventy patients with intra-abdominal infection with non-sporing anaerobes were prospectively studied in an international multicentre study. Patients were randomly allocated to treatment with clindamycin or metronidazole, for a minimum of 48 h to a maximum of 7 days. Other antimicrobial therapy was permitted if indicated by in vitro susceptibility testing. The commonest infections were peritonitis, intra-abdominal abscesses and appendicitis (72 cases), colorectal carcinoma (23 cases), intestinal perforation (16 cases) and diverticulitis (13 cases). Thirty patients received no other antimicrobial chemotherapy and in a further 94 patients, an aminoglycoside was given in addition to the study drugs. In 38 patients the infection required no surgical intervention. Appendicectomy was commonly performed and surgical drainage of pus was required in 14 patients. These variables were evenly distributed between the treatment groups. Both clindamycin and metronidazole were found to be effective therapy for anaerobic infections and were well tolerated. Of the 9 deaths in the study, 7 were in the clindamycin group, and 2 in the metronidazole group. The study protocol allowed patients who were responding poorly to treatment to be crossed over to the alternative therapy. This procedure was followed in 6 patients, 5 of whom were originally receiving clindamycin. It is concluded that metronidazole is as effective for anaerobic infections as clindamycin.