OBJECTIVE Analysis of ascitic fluid is necessary to determine the etiology and to distinguish portal hypertension (PH)-related and unrelated ascites. Numerous diagnostic parameters have been studied, but no single parameter has completely distinguished these. We aimed to validate the serum albumin-ascites gradient (SAAG) for the diagnosis of ascites secondary to PH and to establish cutoff points to predict PH using its sensitivity and specificity. METHODS This was a cross-sectional study conducted on patients diagnosed with ascites of any etiology. The SAAG and albumin concentration in ascitic fluid (AFA) were measured to establish their sensitivity and specificity for determining the presence or absence of PH. Cutoff points and levels of statistical significance were established based on the area under the curve. RESULTS Eighty-seven patients were evaluated, of whom 74 (84%) were men, with an average age of 54.0 ± 13.6 years. Seventy-two (83%) were diagnosed at admission with PH-related ascites and 15 (17%) with non-PH-related ascites. SAAG correctly classified 48 (67%) patients, but 24 (33%) were classified incorrectly, while AFA classified 59 (82%) correctly and only 13 (17%) incorrectly. The diagnostic accuracy of SAAG was 57 versus 73% for AFA. AFA had a sensitivity of 82% and specificity of 66% (95% confidence interval [CI]: 0.63-0.93), while SAAG had a sensitivity of 66% but a specificity of 86% (95% CI: 0.72-0.95). CONCLUSIONS The SAAG showed poor diagnostic performance with low sensitivity but high specificity. The diagnostic accuracy of AFA is superior to that of SAAG in discriminating between PH and non-PH ascites.