A new method for tracking the maneuver of a mobile vehicle by fusing the information of network-based positioning and the local sensors equipped on the vehicle is proposed. We employ the positioning system of the commercial mobile communication network as the network based positioning. The local sensors used here are internal sensors, such as position and velocity sensors of the vehicle. In order to reject random noises in the global and local position sensing, the variable-dimension Kalman filter is employed to fuse the information from different sources and estimate the position of the vehicle. The performance of this new method has been verified by simulations on the tracking of maneuvering vehicles.