The axle-box bearing is a key component of a high-speed train, providing a safety-critical load path for axial and radial forces transmitted between the wheel-rail interface and the bogie frame. Failure of axle-box bearings can directly affect operational and safety performance of the train, and therefore understanding their operating state is key to predicting potential failure modes and planning maintenance interventions. In this study, a three-dimensional vehicle-track coupled dynamics model, with the inclusion of the axle-box bearings has been developed. Based on the bearing's structural properties and operating characteristics, the coupling effects of the vehicle components relative to the bearing are considered. The model also takes a number of non-linear factors into account, such as time-varying bearing stiffness, bearing clearance, wheel-polygonal wear and the non-linear wheel-rail contact forces. Dynamic analysis of the model has been carried out through numerical simulations that consider different amplitudes and harmonic orders of measured and idealised polygonal wheel wear. Both model analysis and validation was supported by field tests performed on a high-speed rail line. Simulation results show that the amplitude of rolling contact forces of the axle-box bearing increases with vehicle speed and the amplitude of polygonal wear. However, the influence of wear upon the axle-box bearing is found to be small in the low-speed range, but at higher operating speeds, polygonal wear begins to lead to a rapid increase in bearing contact forces and the potential for degradation. As a result of vehicle operations, both high order (17 th to 20 th order) and lower order (1 st to 4 th order) wheel polygonal wear patterns are generated. The former's influence on axle-box bearing forces is more significant. Hence, the identification and rectification of this order of wear should form part of maintenance practice and potentially also be considered in the design of axlebox bearings for high-speed trains.