OBJECTIVES: The newly released eighth edition of the American Joint Committee on Cancer TNM staging system for lung cancer seeks to improve prognostic accuracy but lacks external validation for small cell lung cancer (SCLC). Moreover, previous studies posed a few questions concerning survival differences for patients with specific site N3 node involvement or single-site metastasis (SSM) in different distant organs. The aim of this study was to validate the eighth edition of the TNM classification for SCLC in an independent multi-institutional cohort from China and answer the questions raised by the previous research. METHODS: Patients with SCLC from four Chinese cancer centers between 2009 and 2019 were reclassified according to the seventh and eighth edition of the TNM classification. Survival was estimated using the Kaplan-Meier method. Comparisons between adjacent categories and stage groups were performed using Cox proportional hazard regression. R2 statistics were calculated to evaluate the discriminating performance of editions. RESULTS: Of 3384 enrolled cases, 3358 had clinical stage, 537 had pathological stage, and 511 had both. Progressive deterioration of survival was observed with advancing of TNM categories and stages both in the seventh and the eighth edition. The eighth edition stages had a higher R2 statistic than the seventh edition (0.207 versus 0.197). Newly defined categories M1b and M1c and stages IIIC, IVA and IVB in the eighth edition discriminated groups with significantly different prognosis. Patients with N3 contralateral supraclavicular nodes had a significantly worse prognosis than those without (p = 0.032). For patients with single-site metastasis, liver involvement showed a worse prognosis compared to brain involvement (p = 0.030). CONCLUSIONS: Our study provided an external validation of the eighth edition of the TNM classification for lung cancer in Chinese patients with SCLC, and confirmed its improved prognostic accuracy compared with the seventh edition. Patients with N3 and M1b might represent heterogeneous populations that warrant further research.