Abstract: A classification-aided track correlation algorithm is presented to improve the correlation performance using features and attributes of the targets in a distributed multisensor information fusion system. First, track correlation problems using kinematic states and classification information of the targets are transformed into corresponding hypothesis tests based on statistical mathematics theory. According to the distinguished factors which are set beforehand, the correlation matrices of kinematic states and classification information are constructed using the statistics of kinematic states and classification information of the targets. Then the fusion track correlation matrix is formed based on the logical “and” fusion rule. Finally, the minimum average statistical distance criterion is used to make a track correlation decision. By Monte Carlo method, the changing disciplines of correct correlation rate with the factors such as the number of correlation information point, the distance between targets in the same formation and the coefficients of the process noise are simulated, using the track correlation algorithm based on target kinematic states and the classification-aided track correlation algorithm. The simulation results show that, in the case of little number of correlation information points, dense targets distribution or large process noise, a better correlation performance can be obtained when the classification information is introduced into track correlation process.