Objective: To investigate the risk for falling, and fall risk related factors by visiting elderly stroke patients in Shanghai Community Health Service Center. Methods: A total of 230 patients aged from 60 to 90 in the Service Center were randomly selected during 2019 using convenience sampling method. Biochemical indicators were detected,and Morse Fall Risk Assessment Scale was used to assess the falling risk of the subjects. According to the short-form mini-nutritional assessment (MNA-SF method), patients were divided into low-risk and high-risk groups. Multivariate Logistic regression model was used to analyze the falling risk factors, and receiver operator characteristic (ROC) curves was used to evaluate the value of risk factors in predicting falling risk in elderly stroke patients. Results: After baseline analysis, Multivariate Logistic regression analysis showed that MNA-SF (OR=0.338, 95%CI: 0.225-0.508, P<0.001), serum albumin(OR=0.513, 95%CI: 0.396-0.664, P<0.001), hemoglobin (OR=0.908, 95%CI: 0.858-0.961, P=0.001), female(OR=4.407, 95%CI: 1.006-19.311, P=0.049), older age(OR=3.464,95%CI: 1.172-10.235, P=0.025) were risk factors for falling in elderly patients. Among these factors, female,older age (≥80 years old) were the risk factors for falling in elderly stroke patients, besides high score of MNA-SF, elevated level of serum albumin were the protection factors, and elevated hemoglobin was less protective. The area under the curve of ROC curve for the model established with 5 risk factors was 0.925 (P<0.05), which has a good performance for predicting falling in elderly stroke patients. Conclusions: It indicates that female, older age (≥80 years old) are the risk factors for falling, while elevated level of serum albumin,hemogolbin, and good nutrition (MNA-SF≥11) are protective factors for falling in elderly stroke patients. Establishment of risk model will help to identify the patients with high risk of falling, which has practical value.