When running a fuel cell system in load following mode, several operating points (i.e. a combination of gas flow rates, pressures, humidities, temperatures and current density) can provide the required net power. Typically, only one of those operating points is exploited in traditional fuel cell system control. This paper introduces a model-based approach to determine and optimise overall fuel cell system efficiency by choosing an appropriate operating point for a given required net system power. This optimum steady-state operating point can be communicated to the control system as a set point value.