Elevated tumor necrosis factor (TNF) levels have been reported in various models of acute and chronic inflammation and used by many investigators to determine the stage of disease and effectiveness of treatment. Because of the documented involvement of TNF in the mechanism of septic shock, experiments were done to determine whether serum TNF levels paralleled the pathology in endotoxic shock and other models of inflammation. When mice received an intraperitoneal injection of lipopolysaccharide, serum TNF levels increased dramatically, peaking 90 minutes after injection. In a dose-response experiment with lipopolysaccharide alone, we found no correlation between serum levels of TNF and survival rate of mice. All three lipopolysaccharide concentrations resulted in comparable elevations of serum TNF, yet only in the high-dose group did the animals die. In a second model of endotoxic shock, TNF-alpha levels in serum were again compared with the survival rate of mice receiving lipopolysaccharide plus galactosamine. As in the first model, we found no relationship between the level of TNF in mouse serum and mouse survival rate. The two lowest concentrations of lipopolysaccharide/galactosamine induced identically low levels of serum TNF, yet in one group all of the animals survived and in the other all died. Discrepancies between serum TNF level and mortality rate were also seen in drug treatment experiments. GI 147404X, a standard phosphodiesterase type IV inhibitor, inhibited lipopolysaccharide/galactosamine-induced elevation of serum TNF by 90% at doses of 1 and 10 mg/kg. However, the high dose resulted in 66% protection while the low dose afforded no protection.(ABSTRACT TRUNCATED AT 250 WORDS)