In this paper, a novel energy management system (EMS) for a hybrid power source in a small urban electric vehicle (EV) is presented. The problem is divided in three layers, according to the real-time needs of the system. The long-term management layer dynamically restricts the search space based on a set of rules. The short-term management layer implements the optimization strategy based on a hybrid metaheuristic approach combining particle swarm optimization (PSO) and simulated annealing (SA). The solutions to the optimal power sharing problem are used to generate the power references for the lower layer, i.e., a high-speed dc-dc controller. This controller is developed using a linear-quadratic regulator (LQR). A reduced-scale prototype has been built to analyze the performance of the energy management strategy. The results show the effectiveness of this multilayer EMS allowing to fulfill the requested performance with better source usage and much lower installed capacities.