Since plasma concentrations of nitrite/nitrate, the stable end-products of nitric oxide, increase in patients with hepatocellular carcinoma (HCC) correlatively to tumor volume, we examined the ability of plasma nitrite/nitrate to discriminate between those patients with HCC and those without and compared the diagnostic performance of the parameter with that of serum alpha-fetoprotein (AFP) concentrations. Plasma nitrite/nitrate and serum AFP concentrations were measured using a Griess reaction and a solid phase enzyme immunoassay, respectively. Eighty-nine patients with chronic liver diseases (CLD) with (n=39) or without HCC (n=50) and 50 healthy control subjects participated in the study. A receiver operating characteristic (ROC) curve was used to determine the optimal cut-off value and accuracy. The areas under ROC curves for nitrite/nitrate and AFP were calculated to be 0.758 and 0.812, respectively, which were not significantly different. There was no correlation between the concentrations of plasma nitrite/nitrate and serum AFP. The sensitivity, the specificity, and diagnostic efficiency were 79.5, 72.0, and 75.3%, respectively, for nitrite/nitrate, and 74.4, 76.0, and 75.3%, respectively, for AFP. Based on a partial ROC curve, the clinical utility of plasma nitrite/nitrate as a tumor marker approximated that of serum AFP, but exceeded in AFP-negative patients. Indeed, nitrite/nitrate was positive in 70% of AFP-negative HCC patients. The simultaneous determinations of serum AFP and plasma nitrite/nitrate concentrations gave significant improvement in detection of HCC in CLD patients compared with that of serum AFP alone.