Vehicle sideslip angle is significant for electronic stability control devices and hard to estimate due to the nonlinear and uncertain vehicle and tire dynamics. In this paper, based on the two track vehicle dynamic model considering the tire pneumatic trail variation, the vehicle sideslip angle estimation method was proposed. First, the extra steering angle of each wheel caused by kinematics and compliance characteristics of the steering system and suspension system was analyzed. The steering angle estimation method was designed. Since the pneumatic trail would vary with different tire slip angle, distances between the center of gravity and front&rear axle also change with the tire slip angle. Then, based on the dynamic pneumatic trail and estimated steering angle, we modified the traditional two track vehicle dynamic model using a brush tire model. This model matches the vehicle dynamics more accurately. In addition, we designed two extended Kalman filters based on the traditional two track vehicle dynamic model and modified two track vehicle dynamic model to estimate the vehicle sideslip angle. In the end, the extended Kalman filters were validated under simulations and tests under slalom-maneuver and double lane change maneuver in high friction proving ground and low friction winter proving ground. The RT3000 and Correvit S-350 non-contact optical sensor were used to measure the actual sideslip angle. The estimation results were compared with the actual sideslip angle which shows the accuracy of extended Kalman filter with modified two track dynamic model is higher than the extended Kalman filter with traditional two track dynamic model.