To develop a predictive model based on the perioperative plasma concentration of amino-terminal pro-brain natriuretic peptide (NT-pro BNP) in elderly patients to assess the risk of severe postoperative complications. Elderly patients (age ≥ 65 years) enrolled in this prospective observational study underwent general surgery. Plasma NT-Pro BNP concentration was measured before surgery and 2 h after surgery. Univariate and multivariate logistic regression analyses were used to identify the significant predictors. To evaluate the model performance, we applied the receiver operating characteristic (ROC) curve, Hosmer-Lemeshow test, and decision curve analysis (DCA) to evaluate the net clinical benefit. Prediction model was visualised by drawing nomogram and establishing web calculator. A total of 174 elderly patients were enrolled; 15 patients (8.6%) developed severe complications. The area under the ROC curve, sensitivity, and specificity of the two prediction models were 0.899 (95% CI 0.845-0.940), 86.67%, 91.82%, 0.956 (95% CI 0.902-0.985), 100%, and 81.42%, respectively. The net benefit of the post-model was higher than pre-model. We established two postoperative severe complication assessment models based on perioperative NT-Pro BNP levels for elderly patients with reliable accuracy. The nomogram and web calculator will be easy to use by clinicians and other researchers.Clinical significance: The biomarker, NT-pro BNP seem to correlate with some postoperative complications, however no studies have evaluated its relationship with severe postoperative complications in elderly patients. In this study, we evaluated the relationship between NT-pro BNP and severe postoperative complications in elderly patients, and established a prediction model and a web calculator based on the prediction model. Clinicians can easily use this prediction model to identify high-risk patients at an early stage.© 2025. The Author(s).