Platelet-related indices, including mean platelet volume (MPV) and plateletocrit (PCT), have been reported as new prognostic factors of overall survival (OS) in many cancers, but not yet in biliary tract cancer (BTC). We intended to assess these indices in predicting OS in BTC patients with the aim to build a new prognostic model for patients with BTC after surgical resection. Materials and methods Survival analysis and time receiver operating characteristic (time-ROC) analysis were applied to screen the platelet indices. Univariate and multivariate Cox analyses were used to identify independent prognostic factors and develop a new prognostic model. Harrell's C-statistics, calibration curves, and decisive curve analysis (DCA) were used to assess the model. Results MPV and platelet distribution width (PDW)/PCT showed the best prognostic accuracy among the platelet indices. In multivariable analysis, factors predictive of poor OS were presence of nodal involvement, Non-radical surgery, poor tumor differentiation, carbohydrate antigen 19-9 > 100 U/mL, MPV > 8.1 fl, and PDW/PCT > 190. The new model was found to be superior to the TNM staging system and our new staging system showed higher discriminative power. Conclusion MPV and PDW/PCT have high prognostic value in BTC patients, and the novel staging system based on these two indices showed good discrimination and accuracy compared with the American Joint Committee on Cancer 7th TNM staging system.