Background Hepatic encephalopathy is a complex neuropsychiatric syndrome seen secondary to acute liver failure, chronic parenchymal liver disease, or portal-systemic anastomosis. Vasodilatation induced by nitric oxide (NO) may be involved in the development of hepatic coma. However, there are no comprehensive data concerning the effects of NO inhibition on the severity of hepatic encephalopathy. Methods Male Sprague-Dawley rats weighing 300-350 g were used. Fulminant hepatic failure was induced by intraperitoneal injection of thioacetamide (TAA, 350 mg kg(-1) day(-1)) for 3 days. Rats were divided into two groups to receive either NG-nitro-L-arginine methyl ester (L-NAME, 20 mg kg(-1) day(-1) via intragastric gavage) or normal saline (N/S) from 2 days prior to TAA administration for 5 days. Severity of encephalopathy was assessed by counts of motor activity and neurobehaviour test scores. Plasma levels of endotoxin, tumour necrosis factor-alpha and nitrate/nitrite were determined by the chromogenic Limulus assay, enzyme-linked immunosorbent assay and colorimetric assay, respectively. Results Compared with N/S-treated rats, the mortality rate was significantly higher in rats receiving L-NAME (59% vs. 18%, P < 0.01). Inhibition of NO had detrimental effects on the counts of motor activities (P < 0.05) and neurobehaviour score (P < 0.01). Rats treated with L-NAME had significantly higher plasma levels of endotoxin (26.7 +/- 3.8 pg mL(-1)) and tumour necrosis factor-<alpha> (29.4 +/- 6.5 pg mL(-1)) compared with rats treated with N/S (13.2 +/- 2.7 pg mL(-1) and 11.2 +/- 2.6 pg mL(-1), respectively, P < 0.01). Plasma levels of endotoxin and tumour necrosis factor-a, but not of nitrate! nitrite, were significantly correlated with the severity of hepatic encephalopathy (P < 0.05). Conclusion Chronic L-NAME administration had detrimental effects on the severity of encephalopathy in TAA-treated rats, suggesting a protective role of NO in the development of fulminant hepatic failure.