Abstract: An improved adaptive genetic algorithm is presented in this paper. It primarily includes two modified measures, one is anovel adaptive probabilities of crossover and mutation, the other is truncated selection method based on individual's fitness. Thenthis algorithm is applied to optimizing thinned retrodirective array. The directions of each element are taken as variables for optimization. After optimization, sparse retrodirective array keeps excellent retrodirectivity, simplifies the array structure effectively. Result shows that the optimized array precedes the results shown by the existing related literatures.