This paper proposes the self-organizing form-generation system that enables to obtain diverse design solutions without an initial form, in order to support design process that requires diverse design solutions. In the past study, cellular automata were expanded into the form-generation method, and the present study, expanded the form-generation method into the system with the evaluation process. Then, genetic algorithm was adopted as a search algorithm, in order to raise the efficiency of form-generation. Finally, the system was applied to a chair design, and as a result, diverse chair forms were obtained. That results indicated the possibility of utilizing the system for an artifacts design.