Increasingly strict emission and fuel economy standards stimulate the researches on hybrid electric vehicle techniques in the automobile industry and one of the most important techniques is the design of powertrain configurations. In this paper, a theoretical design methodology for hybrid electric vehicle powertrain configurations is proposed to find the configurations with excellent performance in a large pool of configurations. There are two main parts in a powertrain configuration, power/coupling devices (engine, electric machine, wheel, and planetary gear set) and mechanical connections between these devices. Different connections will lead to configurations having different performances. This paper divides all connections in configurations into three categories and a novel matrix representation method is developed to express these kinds of connections to reflect system dynamics and physical structure of configurations. With the support of the matrix representation method, configuration selections from large pools can automatically be completed by computer and manual calculation and comparison can be avoided, which saves much energy and time. Finally, the proposed method is vigorously verified by simulations.