In a prospective, randomized, single-blind trial, we studied 112 adults with intra-abdominal infections and compared antibiotic therapy with cefoxitin plus placebo to therapy with tobramycin plus clindamycin. Seventy-five percent of patients receiving tobramycin-clindamycin and 71% of those receiving cefoxitin-placebo had either shock, bacteremia, malnutrition, alcoholism, rapidly or ultimately fatal underlying disease, infection originating from the distal small bowel or colon, or had had failed therapy before treatment ("high-risk" group). One third of the patients in both groups grew bacteria in the initial culture resistant to the antibiotic regimen used. Ten patients receiving cefoxitin-placebo (17%) and 11 receiving tobramycin-clindamycin (21%) had recurrence of infection or died of infection (clinical failures). Nineteen failures occurred in high-risk patients (p less than 0.05) and 17 were in patients that had antibiotic-resistant bacteria in the initial culture (p less than 0.01). Adverse effects were rare and remitted after antibiotics were stopped. Our results suggest that both cefoxitin and tobramycin-clindamycin are appropriate antibiotic regimens to treat intra-abdominal infections. Clinical failure is more common in high-risk patients and when antibiotic-resistant organisms are isolated from initial cultures.