ABSTRACT BACKGROUND - Variceal bleeding has a high mortality among cirrhotics, and screening with endoscopy is indicated at the diagnosis of cirrhosis. Screening with endoscopy implies discomfort, risks and considerable costs. OBJECTIVE - To evaluate platelet count squared/spleen diameter-aspartate aminotransferase ratio (PS/SA), as a non-invasive predictor of esophageal varices in cirrhotics. METHODS - This cross-sectional study evaluated cirrhotics for PS/SA and presence of esophageal varices. Outpatient records of cirrhotic patients were reviewed for the abovementioned data. Sensitivity, specificity, negative and positive predictive values of PS/SA were calculated. After the univariate analysis, variables with P<0.10 were submitted to a logistic regression. RESULTS - The study included 164 cirrhotics, 59.70% male, with a mean age of 56.7 years. Hepatitis C was the most frequent cause of cirrhosis, being present in 90 patients. Patients were classified as Child-Pugh A in 52.44% and as Child-Pugh B or C in 47.56%. Esophageal varices were present in 72.56% of the patients at endoscopy. PS/SA, with a cutoff of 3x108, had a sensitivity of 95.80% (confidence interval of 95% - 95%CI=0.92-0.99), a specificity of 22.70% (95%CI=0.10-0.35), a positive predictive value of 77.20% (95%CI=0.70-0.84) and a negative predictive value of 66.70% (95%CI=0.42-0.91). In the logistic regression, only platelet count and Child-Pugh score were associated to esophageal varices (P<0.05). CONCLUSION - PS/SA has an excellent sensitivity to predict esophageal varices, allowing almost one fourth of patients without esophageal varices to spare endoscopy. Nevertheless, PS/SA is not independently associated to esophageal varices.