The control problems that arise in connection with hybrid powertrains are significant, and pose great challenges to power-train control engineers. Big obstacles to the control design are the model complexity, and the necessity of a knowledge “a priori” of torque and velocity profiles for optimal fuel consumption and exhaust emissions. The aim of this paper is to propose a robust approach to the development of supervisory control strategies for hybrid-electric drivetrains. The objective is to determine an output feedback controller that minimize fuel consumption with respect to a family of possible torque/power input profiles, e.g. urban driving cycles. Sufficient conditions for the existence of a robust controller satisfying stability, L2 gain attenuation between inputs and outputs of interst, and input/output bound constraints are derived by means of a set of linear matrix inequalities (LMIs).