The systemic tumor necrosis factor (TNF) response has been extensively studied during infection. In addition, antibiotics that cause cell-wall lysis have been associated with endotoxinemia and, therefore, could trigger TNF release. We studied the effects of pretreatment with cefoxitin and/or anti-TNF antibody on mortality and early (90 minutes) and delayed (6 hours) serum TNF levels in a murine model of mixed Escherichia coli/Bacteroides fragilis peritonitis. At low and intermediate inocula levels, cefoxitin, but not anti-TNF antibody, prevented death, and low serum TNF levels were noted in all groups. At the highest inoculum level, mortality was uniform in control, cefoxitin, and anti-TNF antibody groups, and a significant elevation in serum TNF levels was seen only at the 6-hour point in animals receiving cefoxitin. The addition of anti-TNF antibody to cefoxitin at this inoculum level abrogated the 6-hour rise in serum TNF levels and reduced mortality to 40%. These results emphasize that the cytokine response in disease is dependent on both the nature of the insult and other forms of therapeutic interventions.