Abstract: An adaptive neural-fuzzy-based multisensor data fusion architecture for target tracking systems is presented. In this architecture, both process noise and measurement noise are modeled as uncorrelated zero-mean Gaussian noise sequences. Adaptive-network-based fuzzy inference systems (ANFIS) are employed to detect and estimate target maneuvers and measurement noise covariance matrices. They are considered as an adaptive mechanism to cooperate with Kalman filters to process multiple sensor data, which are fused by a specific neural network to obtain optimal results. The results of simulation demonstrate this architecture can avoid mistracking effectively by adjusting tracking parameters