Objective To investigate the compliance to rehabilitation training and its influencing factors in elderly patients after total knee arthroplasty (TKA). Methods The clinical data of 244 elderly patients undergoing TKA in our hospital from January 2022 to January 2023 were collected and retrospectively analyzed. All patients were assessed for rehabilitation training compliance in three months after surgery. The score of training compliance was compared in the patients with different clinical characteristics. Multivariate logistic stepwise regression analysis was used to analyze the risk factors affecting the compliance in the patients, and the obtained risk factors were employed to construct a risk prediction model. Receiver operating characteristics (ROC) curve was plotted to analyze the value of risk prediction models for rehabilitation compliance in the patients. SPSS statistics 22.0 was used to process the data, and student′s t test and F test were applied to compare the data between groups depending on data type. Results For the 244 elderly patients after TKA, the score of physical exercise compliance, active exercise seeking compliance and exercise supervision compliance was (13.27±3.09), (6.88±2.04), and (6.12±1.81) points, respectively. The total score of compliance to rehabilitation training was (23.27±6.94) points, with the lowest score of 15 and the highest one of 32. The total score of rehabilitation training compliance of the patients with female, aged ≥80 years, combined diseases >2, junior high school education level, no spouse, living alone, monthly income <500 Yuan and chronic pain was significantly lower than those of male, aged <80 years, combined diseases ≤2 or no, high school education level or above, having spouse, cohabitation with spouse or children or others, monthly income ≥500 Yuan, and no chronic pain (P<0.05). Multivariate logistic regression analysis confirmed that female (OR=3.544,95%CI 1.036-12.123), aged ≥80 years (OR=1.052,95%CI 1.024-1.081), combined diseases >2 (OR=4.418,95%CI 1.274-15.321), junior high school education (OR=4.274,95%CI 1.542-11.846), no spouse (OR=3.245,95%CI 1.244-8.465), living alone (OR=1.226,95%CI 1.105-1.360), monthly income <500 Yuan (OR=2.429,95%CI 1.442-4.092), and chronic pain (OR=2.015,95%CI 1.009-4.024) were risk factors for the compliance to rehabilitation training in elderly TKA patients. ROC analysis indicated that the constructed risk model had an area under the curve of 0.934, standard error of 0.014,95%CI of 0.907-0.961, best cut off value of 35.499, sensitivity of 0.925, specificity of 0.890 for predicting rehabilitation training compliance in elderly TKA patients. Conclusion Many factors affect the compliance to postoperative rehabilitation training in elderly TKA patients. ROC analysis confirms that our risk prediction model has a good predictive value for the compliance, which may be helpful for implementation of subsequent rehabilitation training for these patients.