Taguchi``s parameter design is for identifying settings of product/process design parameters that minimize the expected quadratic loss(MSE). Yet, in practice, Taguchi chooses design settings that maximize the signal-to-noise(SN) ratio among the experimental points. In this thesis, a mathematical model and solution method are presented for the parameter design problem in which all design parameters are continuous, adjustment factors may not exist, and general constraints may be considered. The problem under investigation is stochastic and nonlinear in nature due to the random noise in the quality characteristic. The proposed method considers MSE directly, and iteratively searches for an optimal setting within an experimental region using the Complex Search Method coupled with the response surface methodology. Statistical methods for comparing MSE``s are also developed.