We tested the hypothesis that the metabolic changes in glucose, lipid, and protein metabolism seen during tumor necrosis factor (TNF) infusion were due to the increase in plasma glucagon concentration rather than to the direct effects of TNF. We employed a pancreatic clamp technique to keep plasma insulin and glucagon concentrations constant throughout a 4-h isotope infusion. Glucose, lipid, and protein kinetics were measured by means of the primed, constant infusion of [6,6-2H]glucose, [2H5]glycerol, [2H2]palmitic acid, and [1-13C]leucine. After a 2-h baseline period (period 1), TNF was infused as a primed, constant infusion (prime, 2.5 micrograms/kg; constant infusion, 62.5 ng.kg-1.min-1) for 2 h (period 2). Whereas plasma glucose concentration dropped significantly during TNF infusion, endogenous glucose production did not change. The decrease in glucose concentrations was due to a stimulation of glucose clearance (P < 0.05). The rate of lipolysis did not decrease significantly, but free fatty acid (FFA) flux and plasma FFA concentration significantly decreased during TNF infusion (P < 0.05). The rate of appearance of leucine was not affected by TNF infusion, but TNF caused a significant increase in 13CO2 excretion (P < 0.05) and leucine oxidation (P < 0.05). The calculated rates of whole body protein synthesis decreased. We concluded that TNF did not directly affect glucose production. Furthermore, changes in protein and lipid kinetics during TNF infusion were not mediated by changes in insulin or glucagon and may have reflected direct effects of TNF.