Abstract: Making use of the theory of maximum likelihood estimation (MLE), a new multipath estimation method based on colored noise Kalman filter to estimate GPS multipath signal is put forward. A linear model of correlation outputs for multipath estimation based on Kalman filtering was built. The state equation and observation equation of Kalman filter were deduced when the observed noise was colored noise, and the generation process of colored noise was explained. The recursive formula of Kalman filter was deduced, which was used to execute iterative filtering of MLE output. Experimental results show that the new design has better performance than MLE, and it can get the precise estimation of multipath signal under 20dB·Hz of CNR.