Abstract: OBJECTIVE To observe and compare the efficiencies of Nomogram and the Classification and Regression Tree (CART) model in prediction of poor prognosis of the patients with sepsis so as to provide more precise risk prediction tool for clinical decision. METHODS The clinical data were retrospectively collected from 130 patients with sepsis who were treated in Nanyang First People's Hospital from Jan. 2023 to Jan. 2024. The enrolled patients were divided into the survival group with 81 cases and the death group with 49 cases according to the survival status within 28 days after the admission. Univariate analysis and multivariate logistic regression analysis were performed to screen out the predictive factors that affected the poor prognosis of the sepsis patients, the Nomogram model and the CART model were respectively constructed, and the predictive efficiencies of the two models were evaluated by receiver operating characteristic (ROC) curves. RESULTS There were significant differences in C-reactive protein (CRP), procalcitonin (PCT), urea nitrogen, serum creatinine, prealbumin, total bilirubin, APACHEⅡ score and SOFA between the death group and the survival group (P＜0.05). Multivariate logistic regression analysis showed that the rises of CRP, PCT, APACHEⅡ score and SOFA were associated with the poor prognosis of the sepsis patients (P＜0.05). The areas under the ROC curves (AUCs) of the Nomogram model and the CART model were 0.870 and 0.951, respectively; the sensitivities were 86.8% and 90.6%, respectively; the specificities were 86.0% and 80.0%, respectively; the AUC of the CART decision-making tree model was greater than that of the Nomogram model(Z=4.458,P=0.025). CONCLUSIONS Both the Nomogram model and the CART decision-making tree model can effectively predict the risk of poor prognosis of the sepsis patients, and the predictive efficiency of the CART decision-making tree model is higher than that of the Nomogram model.