This thesis is concerned with a problem in which optimal parameter settings are determined to make product and manufacturing process robust to uncontrollable variations. To determine an optimal parameter setting an emperical procedure is proposed. The proposed procedure utilizes the concept of the PERMIA(Performance Measure Independent of Adjustment) and approximate expected loss, and can be applied in two distinct cases: one with the prior knowledges on the transfer function and the characteristics of design parameters, and the other without such prior knowledge. In this sense the proposed empirical procedure unifies the previous methods for parameter design. A numerical example is also presented to illustrate the proposed procedure.