Abstract: A co-localization method for lunar rover positioning based on very long baseline interferometry (VLBI) and celestial navigation systems (CNS) is presented in the paper. A federated kalman filter was utilized to implement optimal estimation of position information in order to enhance reliability and fault tolerance ability of the system. Experimentalresults calculated with measurement data from the Chang'E-3(CE-3) mission demonstrate that this method improves the positioning accuracy of the lunar rover comparing with VLBI alone or joint calculation with Least Squares method. Furthermore, this method also guarantees the reliability and stability of lunar rover positioning.