Based on a brief review of the developing history of vehicle handling inverse dynamics, completely considering the cornering force of tires, the nonlinear vehicle model of two-degree-freedom was established with steering angle input. Using radial basis function neural networks, the nonlinear mapping relation between lateral angular velocity and steering angle input was founded with different friction coefficient at high velocity. The identification results of step steering angle input showed that the model is not only feasible, but also have high accuracy, good stability and better computational speed.