Ninety patients infected with presumed penicillin resistant anaerobes were randomized to cefoxitin or clindamycin-aminoglycoside. Cefoxitin was comparable to clindamycin-aminoglycoside in cures of intestinal associated, 16 of 26 vs. 11 of 21, and pelvic infections, 20 of 20 vs. 22 of 23. Cefoxitin-resistant facultative-aerobic gram-negative rods were found in 16 of 45 patients with intestine associated infection. Probable antibiotic associated nephrotoxicity was less frequent in the patients in the cefoxitin group, 0 of 46 vs. 7 of 44, P < 0.05, although a false creatinine elevation was noted more frequently, 7 of 46 vs. 1 of 44, P < 0.05. Infections causing failure in patients in the cefoxitin group more frequently contained cefoxitin resistant gram-negative rods at the time of failure than did infections causing failure in those in the clindamycin-aminoglycoside group that contained gentamicin-resistant gram-negative rods, 8 of 8 vs. 0 of 8 P < 0.001. Cefoxitin may be adequate therapy for many patients with mixed anaerobic/aerobic infections; the addition of an aminoglycoside may be prudent in those with known, or suspected, cefoxitin resistant gram-negative rods.