In this paper, an improved selection method is proposed and integrated with summation of normalized objectives based multi-objective differential evolution to solve multi-objective optimization problems. The summation of normalized objectives and diversified selection is used to replace the non-domination sorting and reduce the simulation time of the multi-objective evolutionary algorithm. However, the diversified selection may keep some bad individuals as parents which lead to poor performance. With the proposed method, a pre-selection is applied to filter the bad solutions and improve the convergence. The algorithm is tested on 15 commonly used benchmark problems and compared with a number of multi-objective evolutionary algorithms in literature. The results show that the proposed algorithm is effective and efficient.