Designs of machine products routinely have so many characteristics to be evaluated that usual design optimization methods often result in an unsatisfactory local optimum solution. In order to overcome this problem, this paper proposes a design optimization method based on decomposition by substructuralization and subsequent hierarchical ordering, considering the both conflicting and cooperative relationships between the characteristics under evaluation. First of all, each characteristic is divided into simpler basic characteristics. The pool of design variables is also divided into smaller groups, according to specific design features. Next, the relationships between the basic characteristics and the divided design variables, as well as the relationships among the characteristics themselves, are systematically identified and clarified. Then, based on this clarification, and after setting a core characteristic derived from the primary performance characteristic for the product under consideration, an optimization strategy and detailed hierarchical optimization procedures are constructed. In this paper, the proposed method is applied to machine tool structures and transportation products.