The bat algorithm is introduced to solve the NP problem in multi-sensor multi-target allocation. By applying the method of K-means for initializing the algorithm, adopting the adaptive step length for updating the speed, searching in the opposite direction and dissociating, the basic bat algorithm is improved. A simulation is made and the result shows that: 1) Compared with the basic bat algorithm, the improved bat algorithm has faster calculation speed and better optimizing capability; and 2) Compared with the particle swarm algorithm, the artificial bee colony algorithm, and the wolf algorithm, the improved bat algorithm is the most adaptive to the problem of multi-sensor multi-target allocation, and can obtain the best solution.