Abstract: Based on particle filtering(PF),unscented particle filter(UPF)was proposed for satellite attitude estimation based on modified rodrigues parameters(MRPs) and dynamic equation for attitude representation and the propagation of angular velocity respectively.The unscented kalman filter(UKF) was used to generate sophisticated proposal distributions,which can take advantage of the latest measurements.The simulation results show that the presented method has higher accuracy than that of the standard PF and UKF.