We infused recombinant human tumor necrosis factor alpha (rhTNFJ, lymphotoxin (rhLT), and Escherichia coli 0111:B4 lipopolysaccharide (LPS) into anesthetized sheep with a lung lymph fistula to compare their effects on systemic and pulmonary bemodynamics, lung lymph dynamics, and eicosanoid release. rhTNF,, (25-150 pg/kg, n-6 sheep), but not rhLT (25 /ig/kg, n=3), rapidly increased lung lymph and plasma levels of 6-kcto-prostaglandin F la (6-k-PGF,,,) and caused profound systemic vasodilation and hypotension. Meclofenamate pretreatment (10 mg/kg) of three other sheep given 25 Mg/kg rhTNF., prevented the increase of lymph and plasma 6-k-PGF la levels, systemic vasodilation, and the early (<2 hrs) but not the late (4-6 hours) hypotension caused by rhTNF,,. LPS (1 /ug/kg, n = l l ) induced a briefer increase of lymph 6-k-PGF ln levels than did rhTNF a while plasma 6-k-PGF,,, levels did not increase. LPS induced more gradual hypotension than did rhTNF a but did not cause systemic vasodilation. LPS and rhTNF^ but not rhLT, increased lymph thromboxane Bj (TXBj) levels during the first hour of study, whereas only LPS acutely increased plasma TXB 2 levels. LPS caused acute pulmonary vasoconstriction and greater acute pulmonary artery hypertension than did either rhTNF,, or rhLT. Whereas LPS-treated sheep required less fluid transfusion than rhTNF a -treated sheep to maintain mean systemic arterial pressure greater than 50 mm Hg, LPS infusion caused a greater increase of lung lymph protein clearance. rhTNF,, caused minimal alterations of lung microvascular permeability. We conclude that eicosanoid mediators contribute importantly to differences of systemic and pulmonary hemodynamics caused by these agents in sheep.