Abstract In order to significantly reduce vehicular fuel consumption and emission pollutants, power-split hybrid electric vehicles are increasingly deployed to ensure that internal combustion engines work at their high-efficiency regions. Because of multiple power components (internal combustion engine, two electric machines, and a driveshaft to the wheels), configurations of such vehicular powertrains are typically very complicated. In order to systematically analyze and design a fuel-optimal powertrain, an innovative hierarchical topological graph approach is proposed. This method comprises four major design processes: (1) modeling of hybrid vehicle powertrain systems, (2) generation of a configuration pool, (3) identification of isomorphism, and (4) classification of configuration modes. Potential power-split hybrid vehicle designs are rigorously examined via a dynamic programming algorithm to estimate their acceleration performance (0–100 km/h) and fuel economy in various driving cycles.