Abstract: Genetic algorithm is applied to synthesize low sidelobe antenna array. By constructing proper fitness function, computational complexity can be reduced. The searching ability of genetic algorithm is improved by nonlinar transformation of fitness function. Because of genetic algorithm's inherent characteristics, such as implicit parallelism, robustness and simplicity, computational efficiency can be improved, and global optimum results can also be found. The convergence of genetic algorithm is stable, and good results are achieved.