Objective: To evaluate the accuracy of the new non-invasive liver disease model platelet-albumin-bilirubin index (PALBI) in the diagnosis of patients with acute upper gastrointestinal hemorrhage (AUGIB) due to cirrhosis. Methods: 277 patients with AUGIB due to cirrhosis were analyzed retrospectively. The data of platelet, total bilirubin, albumin, creatinine, international standardized ratio and etiology of cirrhosis were collected. Univariate and multivariate logistic regression analysis was used to identify independent risk factors for death in patients with cirrhosis complicated by AUGIB. Analysis of variance was used to compare the differences between the model for end-stage liver disease (MELD) and PALBI grades. Pearson correlation analysis was used to assess the association between MELD and PALBI. The operating characteristic curve (ROC) was used to compare the predictive power of both for short-term and long-term mortality in patients with cirrhosis complicated by AUGIB. Results: The short-term and long-term mortality rates of patients with cirrhosis complicated by AUGIB were 13.7% and 23.5%, respectively. The average hospital stay was (9.1±3.9)days. The high MELD score and high PALBI index were confirmed as independent risk factors of death by single factor and multiple factors [odds ratio (OR)=1.17, 4.43; P<0.05]; the Pearson correlation analysis showed there was a positive correlation between MELD score and PALBI index (r=0.735, P<0.05). The PALBI score was scored in MELD-a patients, further subdivided into PALBI-1a and PALBI-1b. There were statistical difference in the 1-year mortality rate between the two groups (7.0% vs 17.8%, χ²=4.033, P<0.05). The ROC curve was used to compare the predictive power of MELD and PALBI for short-term mortality. The area under curve (AUC) of PALBI was 0.767 (95% CI: 0.712-0.815), while the AUC of the MELD score was 0.651 (95% CI: 0.591-0.707), with statistically significant difference (Z=2.328, P<0.05). The predictive power of PALBI and MELD for long-term mortality were 0.731(95% CI: 0.674-0.782), 0.754 (95% CI: 0.699-0.804), but the difference was not statistically significant (Z=0.828, P>0.05). Conclusions PALBI has a better predictive effect on patients with cirrhosis complicated by AUGIB than MELD scores. PALBI can achieve a more precise prognosis classification for patients with MELD-a, and maintain a good prediction ability on the short-term (within 30 days of hospitalization and discharge) and long-term (within 1 year after discharge) mortality of patients. As a new liver disease model, PALBI can be used as an effective non-invasive means to judge the prognosis of patients with liver cirrhosis complicated by AUGIB .