This paper presents a novel design of kinematics-based observer for vehicle sideslip angle estimation. A parameter-varying model is firstly obtained by regarding the measured yaw rate as a varying parameter. A yaw rate scheduled parameter-varying observer is then constructed using the measured yaw rate and accelerations, and is designed to be robust against the measurement noises. The conditions for designing such an observer are derived in terms of linear matrix inequalities (LMIs). Numerical simulations on a nonlinear 8-DOF vehicle dynamics model and experimental data are used to validate the effectiveness of the proposed approach. The results show that the designed observer can effectively estimate the vehicle sideslip angle regardless of the measurements with noises.