In the design optimization of electric machine, there is a strong need to comprehend in detail the tradeoff relationships among the various objective functions. Therefore, it is important to obtain the sufficiently diverse pareto solutions for appropriately designing electric machine. However, the conventional genetic algorithm (GA) doesn't necessarily find out the diverse pareto solutions. In this paper, we propose a GA with new concept of crowding distance which enables us to obtain the sufficiently diverse pareto solution. Some numerical examples which demonstrate the validity of the proposed method is presented.