Mechatronic design optimization is a complex process characterized by an important number of requirements, design variables, constraints and objectives. Therefore, it is very important to decompose efficiently the system design problem into a set of partitions to minimize the computational cost while profiting from the spatial distribution of design tools, working teams and expertise. However, the optimization of the overall design requires incorporating the relevant partitions in order to find the optimum mechatronic design. Efficient strategies of partitioning and coordination should be specified at the conceptual level to have a successful optimization process. In this paper, a new approach based on multi-agent paradigm is proposed for mechatronic design optimization. The proposed approach is applied to the preliminary design case of an electric vehicle to demonstrate its validity and effectiveness.