A randomized, double-blind study of cefoxitin (CX) or clindamycin/gentamicin (CG) as adjuncts to the surgical management of peritonitis is reported. Groups with similar infection risks were evaluated by including only patients with abdominal stab wounds, enteric injury, and spillage of the gastrointestinal contents. One hundred ninety-five patients were entered of whom 75 were evaluable. Comparisons of the ages, sex, diagnoses, and measures of outcome were not significantly different. Fifteen per cent (5/34) of CX treated patients had postoperative complications (three infections) vs. 10% (4/41) of patients treated with CG (three infections). Intraperitoneal bacteria were cultured from 62% of CX and 59% of CG patients. Antibiotic resistance, seen in three patients of each group, was not associated with failure. Two moderately sensitive Bacteroides distasonis were each associated with a failure in the CX and CG groups. We deduce that both regimens are effective and that cefoxitin may represent less costly single-agent therapy.