AIM: To identify clinical parameters, and develop an Upper Gastrointesinal Bleeding (UGIB) Etiology Score for predicting the types of UGIB and validate the score.METHODS: Patients with UGIB who underwent endoscopy within 72 h were enrolled. Clinical and basic laboratory parameters were prospectively collected.Predictive factors for the types of UGIB were identified by univariate and multivariate analyses and were used to generate the UGIB Etiology Score. The best cutoff of the score was defined from the receiver operating curve and prospectively validated in another set of patients with UGIB.RESULTS: Among 261 patients with UGIB, 47 (18%) had variceal and 214 (82%) had non-variceal bleeding.Univariate analysis identified 27 distinct parameters significantly associated with the types of UGIB. Logistic regression analysis identified only 3 independent factors for predicting variceal bleeding;previous diagnosis of cirrhosis or signs of chronic liver disease (OR 22.4, 95% CI 8.3-60.4, P < 0.001), red vomitus (OR4.6, 95% CI 1.8-11.9, P = 0.02), and red nasogastric (NG) aspirate (OR 3.3, 95% CI 1.3-8.3, P = 0.011).The UGIB Etiology Score was calculated from (3.1 x previous diagnosis of cirrhosis or signs of chronic liver disease) + (1.5 × red vornitus) + (1.2 × red NG aspirate), when 1 and 0 are used for the presence and absence of each factor, respectively. Using a cutoff ≥ 3.1, the sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV) in predicting variceal bleeding were 85%, 81%,82%, 50%, and 96%, respectively. The score was prospectively validated in another set of 195 UGIB cases (46 variceal and 149 non-variceal bleeding). The PPV and NPV of a score ≥3.1 for variceal bleeding were 79% and 97%, respectively.CONCLUSION: The UGIB Etiology Score, composed of 3 parameters, using a cutoff ≥ 3.1 accurately predicted variceal bleeding and may help to guide the choice of initial therapy for UGIB before endoscopy.