A mutation ant colony algorithm for the asynchronous muti-sensor track correlation is proposed.The time synchronization of the asynchronous tracks is implemented by optimal prediction.The infeasible solutions is avoided by dynamically deleting the index of the accessible nodes.The mutation operations for the optimal result is employed to elevate convergent speed and reduce the solving time.A non-uniformly distributional strategy of the initial pheromones is adopted to reduce the invalid assignments.The simulation results show that the computational cost of the algorithm is small and its correct percent of track correlation is iigh.