No laboratory test completely distinguishes malignant ascites (MA) from ascites associated with cirrhosis and (or) hepatocellular carcinoma (A/C-HC). Ascitic cytology is highly specific but has a diagnostic sensitivity of only 40-60%. We determined 11 ascitic analytes and cytology in 58 patients with cirrhosis, 15 with hepatocellular carcinoma, and 21 with MA (10 ovarian cancers, 4 mesotheliomas, 6 gastrointestinal neoplasias, 1 leukemia). Ascitic total protein, cholesterol, pseudouridine, and lactate dehydrogenase (LD), and the ascitic:serum ratios of total protein and of LD showed the most significant differences between the two groups of patients. Stepwise multiple linear discriminant analysis (applying the Wilks' lambda criterion) of several variables, corroborated by the "jack-knife" reallocation procedure, showed that the ascitic cholesterol and ascitic LD association correctly identified 100% of MA and A/C-HC; cytology had a diagnostic specificity of 100%, but identified only 48% of MA. This association may represent a primary tool for the discrimination of ascites of unknown origin, particularly in the presence of negative cytology findings.