The inverted pendulum system has been researched as an experiment target to verify effectiveness of nonlinear control theory. The controller of inverted pendulum has many control parameters. In general , the control parameters are selected by the trial and error approach. However, it is not efficient because the approach is simple sensation. Recently, control parameters are selected by general Genetic Algorithm (GA) approach. However, it is difficult to implement the real system, because the robustness is not considered. In this paper, a method for selecting control parameters of cart-type inverted pendulum by using Genetic Algorithm with a Robust Solution Searching Scheme (GA/RS3) is proposed. The merit of the proposed method is the efficiency in choosing robust solution. Further, the effectiveness of the proposed method is verified by numerical simulation and experimental results.