To control the system radiation risk and track the maneuvering target in clutter, a non-myopic multi-sensor scheduling policy for maneuvering target tracking is proposed. Firstly, the maneuvering target tracking accuracy in clutter was estimated by the interacting multiple model and probability data association (IMMPDA) algorithm. Secondly, the radiation cost was quantized by the emission level impact (ELI) and the posterior carmér-rao lower bound (PCRLB) was utilized to represent the target tracking performance. Then the radiation cost and the PCRLB over the future finite time horizon were predicted, respectively. Finally, considering the switching cost and the tracking accuracy constraint, a non-myopic multi-sensor scheduling policy with cost function and PCRLB was set up. The constrained scheduling problem was converted to a decision optimization problem which can be solved by a search algorithm with threshold pruning technique. Simulation results show the effectiveness of the proposed policy. Compared with the uniform cost search (UCS), the proposed search algorithm can reduce the number of nodes opened and improve the search speed with a slightly increased radiation cost. Compared with random scheduling, closest scheduling, and greedy scheduling, the proposed policy can obtain lower radiation cost while satisfying the target tracking requirement. Furthermore, the proposed policy also has the lowest switching cost, and the sensor scheduling frequency has been reduced effectively.