Serum and plasma from horses injected with endotoxin was examined for cytotoxic activity. Each of the cell lines, L929 and WEHI 164 clone 13, was sensitive to the cytotoxic effects of equine serum; however, a precipitation artifact caused by the use of isopropanol in the WEHI assay limited the use of this assay to samples containing less than 2 mg of protein/ml. In foals treated with a sublethal IV bolus of 5 micrograms of lipopolysaccharide (LPS)/kg and in adult horses given a low-dose continuous infusion of LPS (30 ng/kg/h for 4 hours), cytotoxic activity was detected in all serum or plasma samples taken between 30 minutes and 4 hours after LPS infusion began. In horses given either continuous or bolus LPS infusions, circulating cytotoxic activity peaked at 1 to 2 hours before decreasing sharply. The onset of pyrexia after LPS infusion coincided with the appearance of circulating cytotoxic activity, but the temperature remained high, even after cytotoxic activity disappeared. Treatment of horses with flunixin meglumine (1 mg/kg) appeared to blunt the pyrexic effect of low-dose continuous LPS infusion, but had no significant effect on circulating cytotoxic activity. Incubation of serum samples with an antibody raised against a portion of human tumor necrosis factor (TNF) resulted in the removal of greater than 90% of serum cytotoxicity, suggesting strongly that the cytotoxic activity was attributable to TNF. These findings are consistent with the hypothesis that TNF is an early acting mediator of the effects of endotoxin in the horse.