: We present a new distributed mechanism for tracking moving objects with a network of sensors. To track such objects efficiently and accurately, we need techniques that allow the cooperation of many sensors, and the real-time exchange of data in the presence of possible failures and within the constraints of the system. Our mechanism provides efficient set-up and cooperation of the sensors within the network, while providing fault tolerant characteristics through replication. We provide an algorithm for predicting, with high probability, the future location of an object based on past observations by many sensors. We empirically evaluate the efficiency and accuracy of our approach.