Quickest and accurate maneuver detection is critical to modern tracking systems. In this paper, the target maneuver detection problem when using multiple sensors is investigated. The target dynamic model and measurement model may exhibit complex nonlinearity and non-Gaussianity. Therefore, particle filters are implemented at the local sensors to predict the target state. At each time step, local sensors transmit binary data to the fusion center, where decision fusion is performed to detect the potential occurrence of target maneuver. Since the sensors observe the same dynamic process, their measurements, and thus the local decisions, are correlated, which has to be taken into account at the fusion center. By considering correlation and using the Bahadur-Lazarsfeld expansion in the fusion rule, we can achieve better system design (local decision rules and fusion rule) than that achieved by assuming independence between sensors. Experimental results show that the distributed maneuver detection system achieves much better performance than using only a single sensor; the correlated design outperforms the independent design, and is very close to the optimal performance, especially for high correlation scenarios.