If fuel cell technology – with its inherent benefits of high efficiency and low emissions – is to be used in decentralised power sources, in mobile or transportation applications, the systems have to be able to adapt to fast load changes and varying operating conditions. In order to achieve such performance, the balance of plant systems – typically governed by an on-board system controller – need to dynamically supply the fuel cell stack with reactant gases at the right flow rates, pressures and humidities while keeping the fuel cell at its correct operating temperature. Since best overall system performance is achieved by using model-based controllers, an appropriate model is required to implement such controllers. This thesis provides a control-oriented state space model for a PEM fuel cell system. The model describes the effects of a user interaction with any of the balance of plant actuators on overall system performance. The system model is elaborated in a two-step process. In a first step, an analytical, steady state, cell-averaged, isothermal fuel cell stack model is developed. The model predicts the fuel cell voltage and membrane water content as a function of the stack's operating conditions – i.e. reactant flow rates, pressures and humidities as well as cell temperature. It provides an analytical expression to the overall water transport within the fuel cell stack. In the second step, dynamic state space models are developed for the balance of plant systems. They link the effects of the auxiliary systems' actuators to the evolution of the operating conditions for the fuel cell stack. In the context of this thesis, state space models for a non-pressurised air supply subsystem, for a recirculating, pressurised hydrogen supply subsystem and for a liquid cooled thermal management subsystem are elaborated. A dedicated fuel cell test bench has been developed that was used to experimentally validate the proposed models.