A second order sliding mode strategy to control the air supply and oxygen stoichiometry of a fuel cell based generation system is presented. The control design is accomplished from a complete model of a experimental plant that was previously developed by the authors and specially suited for nonlinear control issues. The resulting controller endows the system with enhanced dynamic characteristics and robustness to model uncertainties and external disturbances. Simulations and experimental results are provided, showing the feasibility and reliability of the approach.