Abstract: Using tropical cyclone (TC) data of reforecast experiments by 11 dynamical models from the World Climate Research Program and World Weather Research Program sub-seasonal to seasonal prediction project, this study evaluates skills of the 11 dynamical models for predicting TC genesis and track on sub-seasonal time scale over the Westernnorth Pacific, and compares with a statistical model that was developed by regularized logistic regression. The performances of these dynamical models on predicting TC activities at the climatic, interannual and sub-seasonal time scales are analyzed in this study. Results are as follows. (1) The performance of dynamical models on predicting climatic seasonal cycle of TC activities over the Western North Pacific has a critical impact on sub-seasonal forecast skills. If a dynamical model can well reproduce TC activities at the climatic and interannual time scales, there is an expected skill improvement of TC genesis and track forecast on sub-seasonal time scale by improving the model ability for forecasting intra-seasonal oscillation modulation on TC activities. (2) Most of the dynamical models have a better skill for TC track prediction than that for cyclogenesis prediction at sub-seasonal time scale, implying a lower skill for TC intensity prediction in the models. The lower skill for cyclogenesis prediction hampers the improvement of TC track prediction. Improving predictions of climatic and interannual cyclogenesis could enhance tropical cyclone track forecasts by dynamic models.