Background: Thymoma is a type of rare mediastinal tumor whose clinical characteristics and indicators of prognosis are poorly understood. This single-institution retrospective study aimed to assess the predictive value of tumor, node, metastasis (TNM) staging incorporating tumor size in predicting the risk of thymoma recurrence after resection. Methods: Four binary logistic regression models were developed. Models I and II included median tumor size and TNM stage, respectively. Model III included the above two variables. Model IV was model III containing these two variables and their interaction terms. All models were adjusted for WHO histological type, operational time, and adjuvant therapy. Results: A total of 276 patients with a median age of 51.0, including 21 patients with thymoma recurrence, were included in this study. Models II or III showed a lower -2LogL and higher AUC (0.735 and 0.738 vs. 0.576) with significantly better discrimination than model I, and model III and model II shared similar discrimination. In model III, TNM stage was positively correlated with thymoma recurrence. The recurrence risk of patients with TNM stage IV was significantly higher than those with TNM stage I (OR of 11.03, p = 0.022). No significant correlation between the tumor size and recurrence risk (p = 0.779) and no interaction was found between medium tumor size and TNM stage in model IV. Conclusions: This study suggests that the prediction contribution of the TNM stage combined with tumor size is similar to the TNM stage alone for tumor recurrence in patients with thymoma after surgical resection.