We used the single-pass–perfused rat liver model to study short-term regulation of endotoxin-inducible nitric oxide synthesis by following the release of nitrite and nitrate, the oxidation products of nitric oxide, into the effluent perfusate. In endotoxin-pretreated livers, the basal nitrite + nitrate release was 5.3 ± 1.2 nmol·gm liver−1·min−1. Nitrite and nitrate release was stimulated by L-arginine in a dose-dependent and saturable fashion. Maximal nitrite + nitrate release with 1 mmol/L L-arginine infused to the influent perfusate was 10.2 ± 1.1 nmol·gm liver−1·min−1, with a half-maximal effect at 53 μmol/L L-arginine. In the absence of molecular oxygen, nitric oxide synthesis was inhibited. Ammonium chloride infusion also stimulated nitrite and nitrate release to a maximal rate of 9.2 ± 0.8 nmol·gm liver−1 min−1 with half-maximal effects at 60 μmol/L ammonium chloride. Ammonium chloride-stimulated nitrite and nitrate release was abolished when urea synthesis was inhibited by bicarbonate-free liver perfusion. Citrulline and ornithine (200 μmol/L each) were without effect on nitrite and nitrate release. L-Nitroarginine methyl ester inhibited both, L-arglnine-and ammonium chloride-induced nitrite and nitrate release. Stimulation of nitric oxide synthesis by L-arginine and ammonium chloride addition (1 mmol/L each) was accompanied by a threefold-to-fourfold increase of cyclic GMP release into the effluent perfusate. In livers of endotoxin-pretreated rats the urea production from L-arginine was higher than that in untreated livers, suggesting induction of an L-arginine transport system in hepatocytes by endotoxin. The regulation of hepatic nitric oxide production by physiological concentrations of L-arginine and ammonia in the portal vein may be of importance in cirrhosis. (Hepatology 1994;19:641–647).