Background:The present study aimed to develop and validate an effective predictive model for overall survival (OS) in patients with nasopharyngeal carcinoma (NPC).Methods: 519 patients were retrospectively reviewed in this study. Random forest was proposed to identify significant prognostic factors for OS in NPC patients. Then calibration plot and concordance index(C-index) were used to evaluate the predictive accuracy of the nomogram model.Results: We used random forest to select three most important features, dNLR, HGB, EBV DNA, which were significantly relevant to OS of NPC patients. The C-index of our model for OS were 0.733 (95% CI: 0.673~0.793) and 0.772 (95% CI: 0.691~0.853) in the two cohorts, which were higher than that of TNM stage, treatment, and EBV DNA. Based on the model risk score, patients were divided into low-risk and high-risk two groups. Kaplan-Meier curves showed that two subgroups were significantly associated with OS in the primary cohort and in the validation cohort. The nomogram for OS was established including risk score, TNM stage, EBV DNA in the two cohort. It achieved higher C-index of 0.783(95% CI: 0.730~0.836) than that of the risk score model 0.733 (95% CI: 0.673~0.793) in the primary (P = 0.005).Conclusions:The established risk score model and nomogram resulted in more accurate prognostic prediction for individual patient with NPC.