Abstract: For the problem of multisensor-multitarget tracking in nonlinear system, a novel centralized multisensor unscented filter algorithm based on S-D assignment, SD-CMSUKF, is proposed. In the new algorithm, the association of measurements from each sensor to targets is first implemented according to the generalized S-D assignment technique and the optimal partition can be achieved. Then in the optimal partition, the measurements from the same target are dealt with sequentially in terms of the principle of sequential multisensor joint probabilistic data association algorithm (MSJPDA). Based on these, UKF is used for the propagation of state distribution in nonlinear system and the SD-CMSUKF algorithm is derived. Compared with the MSJPDA/EKF, the accuracy and robustness of the proposed algorithm are improved. Simulation results show the superiority of the new algorithm.