Multidisciplinary optimum designs have been of strong interest lately in many fields. A unified multidisciplinary optimum design method by utilizing the genetic algorithms is presented in this study. The design process is modeled there by a distributed hierarchical system to simulate ordinary design processes. Each level is optimized by genetic algorithms somewhat independently but some information is transmitted to the upper or the lower level. Engineering analysis can also be involved in a certain level. The proposed method is then applied to simple examples of preliminary designs of cars and numerical results show the effectiveness of the method.