Abstract: Aiming at the phased array radar system which applies monopulse technology, we adopted digital weighting for suppressing sidelobe of difference beam. We presented to optimize the subarray weights by using genetic algorithm, and constructed two kinds of fitness functions:weighting vector and pattern parameters. Further, we improved the conventional genetic algorithm usin g only one fitness function, and partitioned the optimization process into two s tages, and adopted weight vector as fitness function for the first stage and pat tern parameters for the second one. The improved approach possesses satisfying s tability, and avoids being constrained on local optimal solution, enhances the c alculation efficiency greatly, and improves the sidelobe suppression effect. The simulation results verify the validity of the proposed approach.