Objective To develop a diagram for predicting malnutrition risk in patients on maintenance hemodialysis (MHD) and to assess the predictive validity. Methods A retrospective study was performed in six hemodialysis centers in Chengdu. A total of 1,355 MHD patients were recruited and divided into a normal nutrition group (n=1,030) and a malnutrition group (n=325) based on the controlling nutritional status score. Sociodemographic, laboratory and hemodialysis features one month before nutritional assessment were collected, then multivariate logistic regression was performed to identify the independent risk factors for malnutrition. Risk stratification and risk diagram were determined by R 4.1.2 software. The risk diagram model was internally validated using bootstrap method. Results The prevalence of malnutrition in MHD patients was 23.99%. Age≥60 years, coexisting diabetes mellitus, Kt/V＜1.2, intradialytic hypotension within past 1 month, and parathyroid hormone (PTH)＞600 pg/mL were risk factors for malnutrition in MHD patients (all P<0.05). Low-, medium-, and high-risk categories were created based on the probabilities of malnutrition of 0.231 and 0.537. The diagram model had good discrimination with an area under the ROC curve at 0.862, sensitivity of 72.80% and specificity of 91.50%. The decision curve analysis showed a satisfying net benefit for thresholds between 10% and 94%. Conclusion The diagram could help nurses easily identify patients at risk of malnutrition in MHD and provide reference for further stratified and classified care.