Background and Aims Pancreatoduodenectomy (PD) is a classic surgical method for treating malignant tumors at the pancreatoduodenal junction and other related diseases. Despite advancements in surgical techniques, the incidence of severe postoperative complications remains high. These complications not only affect the patient's recovery process but also pose life-threatening risks. Therefore, predicting the risk of severe complications after PD is crucial for developing targeted prevention and treatment strategies. Recently, sarcopenia, a condition associated with an increased risk of various postoperative complications, has garnered significant attention. The POSSUM scoring system, widely used for surgical risk assessment, has shown preliminary validation in its predictive efficacy. This study was conducted to identify risk factors for severe complications following PD and to develop a risk prediction model based on sarcopenia combined with POSSUM score to improve the accuracy of predicting severe postoperative complications and provide a scientific basis for clinical decision-making.Methods The clinical data of 79 patients who underwent PD from 2016 to 2023 were retrospectively analyzed. The skeletal muscle index at the third lumbar vertebra was obtained using Slice Omatic software, and sarcopenia was diagnosed based on this index. Postoperative complications were recorded and graded according to the Clavien-Dindo classification, categorizing them into severe complications (≥Ⅲa) and non-severe complications (<Ⅲa). The POSSUM scoring system was used to assess surgical risk, and the receiver operating characteristic (ROC) curve was plotted to evaluate the predictive efficacy of the POSSUM score for severe complications after PD, with the optimal cutoff point determined by the Youden index. Univariate and binary multivariate Logistic regression analyses were conducted to identify independent risk factors for severe postoperative complications. Subsequently, a nomogram risk prediction model was constructed using R language, and its predictive efficacy was comprehensively evaluated using the ROC curve, calibration curve, the Hosmer-Lemeshow goodness-of-fit test, and internal validation of the concordance index.Results Among the 79 patients, 41 had sarcopenia, and 38 did not. The incidence of severe postoperative complications was 27.85%. Significant differences were found between the severe and non-severe complication groups regarding age, sarcopenia, POSSUM score, intraoperative blood loss, preoperative white blood cell count, and preoperative neutrophil count (all P<0.05). Binary Logistic regression analysis showed that sarcopenia, POSSUM score, and intraoperative blood loss were independent risk factors for severe postoperative complications after PD (all P<0.05). The risk prediction model constructed based on these risk factors had an area under the ROC curve (AUC) of 0.895 9. The calibration curve of the prediction model was close to the ideal curve, indicating good predictive accuracy. The Hosmer-Lemeshow goodness-of-fit test also suggested a good fit for the prediction model, and internal validation of the concordance index confirmed the nomogram model's good predictive ability.Conclusion Sarcopenia, POSSUM score, and intraoperative blood loss are independent risk factors for severe postoperative complications after PD. The risk prediction model based on sarcopenia combined with the POSSUM score has high predictive efficacy, providing clinicians with a more accurate risk assessment tool and can help develop individualized prevention and treatment strategies to reduce the incidence of severe postoperative complications following PD.