【Abstract】 Objective To investigate the risk factors for malnutrition in patients with multidrug resistant tuberculosis (MDR-TB), and to construct and validate the risk prediction model. Methods A total of 388 patients with MDR-TB admitted to the Fourth People's Hospital of Nanning from January 2019 to December 2021 were enrolled as the training set, and 116 patients with MDR-TB admitted from January to December 2022 were enrolled as the validation set. The training set was used for the construction of a nomogram model, and the validation set was used for the internal validation. According to their nutritional condition, the patients in the training set were divided into the normal nutrition group (n=154) or the malnutrition group (n=234). Univariate analysis and multivariate logistic regression analysis were used to screen the risk factors for malnutrition in patients with MDR-TB, and a nomogram risk prediction model was constructed. The nomogram model was internally validated by the Bootstrap method and assessed by the receiver operating characteristic (ROC) curve, calibration curve, decision curve analysis (DCA), and clinical impact curve (CIC). Results There were statistically significant differences in age, initial treatmeat/retreatment, the combination of severe pneumonia, involved lung field, the combination of pulmonary cavity, body mass index (BMI), albumin (ALB) level, hemoglobin (Hb) level, and CD4+ T lymphocyte count between the normal nutrition group and the malnutrition group (all P<0.05). Multivariate logistic regression analysis showed that initial treatment/retreatment, the combination of severe pneumonia, BMI, and ALB level were independent influencing factors for malnutrition in MDR-TB patients (all P<0.05). In the training and validation sets, the areas under the ROC curves of the nomogram model were 0.851 (95%CI: 0.813-0.888) and 0.717 (95%CI: 0.608-0.827), respectively. The calibration curves showed that the predicted curves were closely aligned with the calibration curves in both the training and validation sets (the mean absolute errors were 0.023 and 0.065 in the training and validation sets, respectively). The results of DCA and CIC showed that in both the training set and the validation set, the nomogram model had clinical net benefits within a large range of threshold probability (0.05-0.91, 0. 93-1.00; 0.43-0.71, 0.79-0.80), with the highest clinical net benefit of 0.6; in the training set and the validation set, the nomogram model had a high clinical predictive value in the range of 0.50-1.00 threshold probability. Conclusion Initial treatment/retreatment, the combination of severe pneumonia, BMI, and ALB level are independent influencing factors for malnutrition in MDR-TB patients, constructed on which the nomogram model has a good clinical application value in predicting malnutrition in MDR-TB patients.