Plain language summary Chordoma is a rare tumor disease. Due to the slow development of the tumor and the atypical symptoms at the beginning, patients often delay the best time to seek medical attention. The reported total resection rate is 40-50%, the 5-year survival rate is about 65% and the average survival time around 6 years. In this study the information of chordoma patients in the Surveillance, Epidemiology and End Results database was used to analyze the influencing factors of overall survival and cancer-specific survival in chordoma patients, and a predictive model for overall survival and cancer-specific survival in chordoma patients was constructed using these influencing factors. Doctors can use this to evaluate chordoma patients' prognosis and choose the best treatment method. Background: The aim of this study was to develop two predictive models to predict overall survival (OS) and cancer-specific survival (CSS) in chordoma patients. Methods: We searched for independent prognostic factors by using univariate and multivariate Cox regression analyses. The prediction model of OS and CSS of chordoma patients was constructed by using the screened factors. Results: The study enrolled 362 chordoma patients. Cox regression analysis showed that disease stage, age, surgery, marital status and tumor size are independent influencing factors of OS and CSS in chordoma patients. After testing, the prediction model constructed in this study has good performance. Conclusion: Two predictive models were successfully constructed and validated for chordoma patients' OS and CSS.