The effects of a 4 h intraportal infusion of Escherichia coli lipopolysaccharide (LPS, .21 mu g/kg/min) on the release of tumor necrosis factor (TNF) by hepatic and nonhepatic splanchnic tissues was assessed in the chronically catheterized conscious dog (n = 7) using arteriovenous difference techniques. TNF levels were measured using both a WEHI-164 cytotoxicity assay (WEHI) and a h-TNF-alpha EIA kit (ELISA; Biosource, Camarillo, CA). Using WEHI, arterial TNF levels increased from 10 + or - 6 pg/mL to a peak of 4667 + or - 1442 pg/mL 100 min after LPS and fell to 443 + or - 199 pg/mL by 240 min. Using ELISA, arterial TNF levels increased from 5 + or - 5 pg/mL to a peak of 12,234 + or - 2046 pg/mL at 100 min and fell to 3511 + or - 991 pg/mL by 240 min. WEHI could not be used to assess organ TNF release due to excessive assay variability. Based upon ELISA, net hepatic TNF output increased from undetectable release at basal to 23.0 + or - 10.7 ng/kg/min at 60 min and returned toward basal by 240 min (4.7 + or - 3.8 ng/kg/min). Net release of TNF by the nonhepatic splanchnic bed was not observed. One compartment analysis of the arterial TNF response indicated that net release of TNF by the liver accounted for the majority of the increase in the arterial TNF levels. In summary, after intraportal LPS infusion, it was determined that 1) both assays predict similar qualitative TNF response, while the quantitative response differs, 2) the liver is the major site of TNF production, and 3) the nonhepatic splanchnic bed is not a net producer of TNF.