Proton exchange membranes are one of the most promising fuel cell technologies for transportation applications. Considering the aim of transportation applications, a simulation model of the whole fuel cell system is a major milestone. This would lead to the possibility of optimizing the complete vehicle (including all ancillaries, output electrical converter and their dedicated control laws). In a fuel cell system, there is a strong relationship between available electrical power and actual operating conditions: gas conditioning, membrane hydration state, temperature, current set point... Thus, a "minimal behavioral model" of a fuel cell system able to evaluate the output variables and their variations is highly interesting. Artificial neural networks (NN) are a very efficient tool to reach such an aim. In this paper, a proton exchange membrane fuel cell (PEMFC) system neural network model is proposed. It is implemented on Matlab/Simulink(R) software and will be integrated to a complete vehicle powertrain. Thus, it will be possible to carry out the development and the simulation of the control laws in order to drive energy transfers on-board fuel cell vehicles. (C) 2003 Elsevier B.V. All rights reserved.