From 1979 through 1981, 152 patients with penetrating injuries of the intra-abdominal gastrointestinal tract were placed on one of three different perioperative antibiotic regimens in a prospective randomized fashion. The three regimens were A) cefamandole 2 grams every 4 hours, B) cefoxitin 2 grams every 6 hours, and C) ticarcillin 3 grams every 4 hours and tobramycin 1.5 mg/kg every 8 hours. Antibiotics were administered intravenously before and for 48 hours following surgical exploration and repair. The three treatment groups were similar with respect to age, average number of organ injuries, and distribution of organ injuries. Cefoxitin-treated patients experienced uneventful recoveries more often than cefamandole-treated patients (94% vs. 80.3%, p less than 0.05) when the incidence of gram-negative wound infection and intra-abdominal abscess formation was considered, while the number of patients who experienced uneventful recoveries in the ticarcillin-tobramycin group was not statistically different from the other two groups of patients. Bacteroides fragilis was isolated from three of the six abscesses occurring in the cefamandole-treated group, while no anaerobes were isolated from abscesses in patients treated with either of the other two regimens. The results of this study suggest that the most effective perioperative antibiotic regimen for patients with penetrating gastrointestinal wounds should possess activity against both aerobic and anaerobic flora of the bowel.