In order to accurately estimate the road adhesion coefficient under different road conditions and improve the safety and stability of vehicles, a method was proposed to estimate the road adhesion coefficient based on the front and rear wheels speed and braking torque under braking conditions. First, considering the transfer of front and rear axle load, the modelling was completed in Matlab/Simulink software to create a dynamic model of the two-wheel vehicle braking.Secondly, the control target was determined as the ideal and actual slip ratio of the front and rear wheels of the vehicle, and the ideal braking torque sliding mode controller was established. The chattering phenomenon of the sliding mode controller was properly dealt with through the integral switching surface. Finally, the front and rear wheel speed and braking torque were used as input to design the extended state observer, which is used to observe the correlation value of the road adhesion coefficient. The results show that the road adhesion coefficient in all kinds of road conditions can be accurately estimated by the above method, and the extended state observer can resist the external interference and has strong robustness. The successful application of the extended state observer provides the reference for the design of vehicle stability control system.