Abstract Objective This study investigated the prognostic impact of multimodality therapies in locally advanced thymomas. Methods From January 1990 to January 2010, clinicopathological, surgical, and oncological features were retrospectively reviewed in a cohort of 370 Masaoka-Koga stage III thymomas (World Health Organization classification A to B3) collected from 37 institutions. A multivariate Cox proportional hazard model was created to identify independent predictors of overall, cancer-specific (CSS), and relapse-free survivals. Furthermore, a propensity score–matching analysis for exposure to adjuvant (AT) therapy was generated. Results Induction therapy and AT were administered to 88 (24.9%) and 245 (69.4%) patients, respectively. Overall, 5- and 10-year overall survival, CSS, and relapse-free survivals were 82.8%, 88.4%, and 80.0%, and 68.9%, 83.3%, and 71.5%, respectively. At multivariable analysis performed in the matched cohort, AT was confirmed as the strongest predictive factor for overall survival (hazard ratio, 2.83; 95% confidence interval, 0.88-9.12; P  = .08) and CSS (hazard ratio, 4.70; 95% confidence interval, 1.00-22.2; P  = .05). Pathologic T classification (according to International Association for the Study of Lung Cancer and International Thymic Malignancy Interest Group TNM staging proposal) was an independent factor for relapse (hazard ratio, 8.69; 95% confidence interval, 1.08-70.04; P  = .04). When CSS was adjusted for T classification, AT confirmed a significant survival advantage for pT3 tumors ( P  = .04). On the other hand, for thymomas larger than 5 cm, stratifying for tumor size and AT did not affect 5-year CSS ( P  = .17). Conclusions Our results indicate that AT is beneficial for locally advanced thymomas, mainly for specific pathologic features (pT3 or tumor size smaller than 5 cm). Further larger studies are needed to confirm these data.