Abstract: OBJECTIVE To explore the risk factors for developing multidrug-resistant bacteria ventilator-associated pneumonia (VAP) in patients with acute severe stroke, and to establish and validate a clinical risk prediction model. METHODS A total of 198 patients with acute severe stroke admitted to the Intensive Care Unit of Shijiazhuang People's Hospital from Jan. 2023 to Jan. 2025 were selected as the study subjects and included in the modeling group. Logistic regression analysis was used to analyze the risk factors for VAP in these patients and to construct a risk prediction model. Additionally, 39 patients with the same condition treated at a branch hospital of Shijiazhuang People's Hospital were selected as the external validation group. The modeling group data were utilized to analyze relevant factors and establish the risk prediction model, while the validation group data were used to verify the model's efficacy. RESULTS Among the 198 patients in the modeling group, 71 developed VAP multidrug-resistant bacteria infections. Logistic regression analysis revealed that mechanical ventilation duration ≥96 hours (P=0.015, OR=2.285), tracheotomy (P=0.036, OR=2.344), hospital stay ≥15 days (P=0.038, OR=2.247) and hypoproteinemia (P=0.015, OR=2.800) were independent risk factors for VAP multidrug-resistant bacteria infections among patients in the modeling group, while hematocrit (P=0.001, OR=0.897) was a protective factor. The receiver operating characteristic curve was plotted for these five independent factors through R Studio software, yielding an area under the curve (AUC) of 0.765 (95% CI: 0.697-0.833), indicating good discrimination of the prediction model. Internal and external validation results demonstrated good accuracy, high calibration and favorable clinical application value of the model. CONCLUSION The risk prediction model for VAP multidrug-resistant bacteria infection in patients with acute severe stroke constructed in this study demonstrates good prediction ability, providing a reference basis for the early clinical identification of high-risk patients.