Objective To explore the high-risk factors and prevention strategies for rebleeding in patients with upper gastrointestinal hemorrhage (UGIH) treated with endoscopy, and construct a predictive model.Methods 97 patients with UGIH who experienced rebleeding after endoscopic treatment from January 2020 to December 2023 were selected as the observation group, and another 178 patients with UGIH who did not experience rebleeding after endoscopic treatment admitted during the same period were selected as the control group, both groups were followed up for 1 year after endoscopic treatment. Clinical data of the two groups was compared, the high-risk factors for rebleeding after endoscopic treatment in patients with UGIH were analyzed by multivariate Logistic regression analysis, a predictive model was constructed, and the predictive value of the model for rebleeding after endoscopic treatment in patients with UGIH was analyzed by plotting a receiver operator characteristic curve (ROC curve) to analyze.Results The proportions of patients in the observation group with liver cirrhosis, shock, endoscopic active bleeding, Forrest classification of I_a to I_b, level of blood hemoglobin ≤ 90 g/L, and level of blood platelet ≤ 100 × 10⁹/L were 55.67%, 14.43%, 37.11%, 62.89%, 23.71%, and 23.71%, respectively, which were higher than the control group's 41.57%, 2.25%, 18.54%, 44.38%, 3.37%, and 7.87%. The level of serum D-dimer (D-D) of the observation group was higher than that of the control group, and the bleeding volume of the observation group was more than that of the control group, the prothrombin time (PT) of the observation group was longer than that of the control group (P < 0.05). Multivariate Logistic regression analysis showed that: cirrhosis (OR＾ = 2.423, 95%CI: 1.124 ~ 5.224), shock (OR＾ = 6.897, 95%CI: 1.487~31.995), endoscopic active bleeding (OR＾ = 2.604, 95%CI: 1.109 ~ 6.118), Forrest grading of I_a to I_b (OR＾ = 2.494, 95%CI: 1.162 ~ 5.354), level of blood hemoglobin ≤ 90 g/L (OR＾ = 5.270, 95%CI: 1.797~15.442), level of blood platelet ≤ 100 × 10⁹/L (OR＾ = 5.018, 95%CI: 1.733 ~ 14.531), bleeding volume > 189.61 mL (OR＾ = 1.025, 95%CI: 1.016 ~ 1.034), PT > 15.99 s (OR＾ = 1.996, 95%CI: 1.618 ~ 2.460) were both risk factors for rebleeding in UGIH patients after endoscopic treatment (P < 0.05). Regression equation model: logit (P) = -18.551 + cirrhosis × 0.885 + shock × 1.931 + endoscopic active bleeding × 0.957 + Forrest grading × 0.914+level of blood hemoglobin × 1.662 + level of blood platelet × 1.613+bleeding volume × 0.025 + PT × 0.691. The ROC curve for predicting rebleeding in UGIH patients after endoscopic treatment was plotted according to the diagnostic probability logit (P). When logit (P) > 0.30, the 95%CI was 0.891 ~ 0.955, and the diagnostic sensitivity and specificity were 88.66% and 83.15%, respectively. The area under the curve (AUC) value was 0.923.Conclusion The cirrhosis, shock, endoscopic active bleeding, Forrest grade I_a to I_b, level of blood hemoglobin ≤ 90 g/L, level of blood platelet ≤ 100 × 10⁹/L, bleeding volume > 189.61 mL, and PT > 15.99 s are independent risk factors for rebleeding after endoscopic treatment in patients with UGIH. The model constructed based on this has high predictive value, which can be used clinically to provide personalized intervention and treatment for high-risk patients to reduce or avoid the occurrence of rebleeding.