Abstract: Binary code signal have been widely used in various radars due to their simple implementation, but the selection of the binary codes with high compression ratio and low sidelobes is not well solved, because of the difficult processing in mathematics and expensive calculation cost. In this paper, neural computing is introduced into the field of the selection of binary codes, and a new method based on simulated annealing is proposed. The experiments show that the proposed method is able to select the optimal binary codes with much less time than the known methods, furthermore the optimization of selection of the binary codes versus the calculation cost trade-off is easier.