The invention belongs to the field of signal processing, and particularly provides a distributed tracking method for an extended target. According to the method, a heterogeneous network is considered,the number of scattering sources observed on each node obeys Gaussian distribution of different mean values, and the observation noise power of each node is different. The method comprises the following steps: firstly, each node calculates to obtain an intermediate estimated value of the dynamic state and shape characteristic of an extended target by exchanging information with neighbor nodes thereof; and then, each node propagates the intermediate estimation result in the neighborhood, and final estimation results of the dynamic state and the shape characteristics are obtained respectively.The method mainly aims at an extended target, and in a distributed network, each node can estimate dynamic states such as the position, the speed and the acceleration of the target and shape characteristics such as the direction and the size of the target at the same time. Compared with a non-communication method of each node, the method provided by the invention can effectively improve the tracking performance of each node, the tracking performance is close to a centralized method of sending the information of each node to a fusion center for processing, Meanwhile, the risk that the whole system crashes when the fusion center goes wrong can be effectively avoided.