Dr. Genichi Taguchi's method ”on-line feedback quality control system” is took the loss function as a foundation to establish control pattern, which minimizes the total loss via modifying the production drift problem. In the on-line feedback quality control system, parameters must take D0 and u0 as the initial values, as well as the information of cost per measurement, cost per adjustment and in-plant cost of reworking or scrapping a unit that falls outside of tolerance interval, to inference the measurement interval and adjustment limit. But the estimated parameters always changed because of the material undulation, production improvement, lot size, personnel's difference and the control method adjusted. Therefore; this article utilizes the methods of orthogonal array, Signal-to-Noise ratios, ANOVA and multiple regression analysis to discuss the influence of variables change in on-line feedback quality control system.