Most of tasks for vertical surface work in shipyard have been accomplished by human workers. However, such manual work often causes injury to workers, also the production cost becomes high due to increasing individual wage. To cope with the circumstance, shipbuilding companies try to introduce wall-climbing robots for carrying out such kind of tasks. In designing a wall-climbing robot, it is essential to minimize its own weight to improve the performance such as moving speed and power saving. For such purpose. this study proposes a method of optimal design for a wall-climbing robot using a genetic algorithm with multi-objective function. Specifically, the thickness of the robot base is minimized to reduce the weight while maintaining the allowable strength and avoiding the resonance frequencies. The proposed method is applied to the design of a wall-climbing robot, and the result shows that the method is useful at an early design stage.