For PEM fuel cells supplied with air, pressure and flow control is a key requirement for an efficient and dynamic operation because fuel cells are in risk of starvation when the partial pressure of oxygen at the cathode falls below a critical level. To avoid oxygen starvation and, at the same time, to allow for a dynamic operation of the fuel cell system, both excess ratio of oxygen and cathode pressure need to be adjusted rapidly. In this paper a model-based control structure is proposed that comprises a multivariable control of the cathode pressure pc and the excess ratio of oxygen lambda(o2) using the mass flow controller (MFC) and the outlet throttle as actuators. For both controlled variables appropriate desired values can be specified independently. As the derived dynamic model of the air supply system is nonlinear, advanced control techniques using differential flatness are applied. Since the partial pressure of oxygen in the cathode flow field is a state variable that is hardly accessible by measurement, a tracking observer is employed to estimate this pressure using the measured air pressure at the outlet throttle. Thereby, the observer enables the control system to detect both temporary shortages of oxygen and peak pressures at the cathode and to counteract accordingly. (c) 2007 Elsevier B.V. All rights reserved.