This paper is intended as a development of a new quality design method of improving quality of products. Conventional quality design methods are for single characteristic and aim at minimizing quality costs. Optimal levels of design factors cannot be determined if there are 110 signal factors. They can be outside feasible regions even if there are signal factors. The proposed method minimizing manufacturing costs is applicable to products with multi objective characteristics and with or without signal factors. A generalized loss function and its statistical expectation for multi objective characteristics are introduced. Classified objective functions are presented considering properties of quality design problems. A minimization problem of the objective function is formulated as a nonlinear programming with a response surface methodology and a central composite design. The effectiveness of the proposed quality design method is shown by numerical examples.