Today the world is full of time-dependent phenomena in all fields: physics, chemistry, mechanics and many others. Time acts on the performance of any system whatever its nature is. When a system operates over time, aging becomes a real concern. Regarding fuel cells, several degradation phenomena can occur in a short or long term. Short-term phenomena are generally referred to as reversible degradations, these degradations are of the order of the microsecond and can go up to hours and sometimes to days, such as problems related to water management. The long term degradations are usually called irreversible degradations; it can be defined as aging. This phenomenon is of the order of a day and can increase up to months. In order to mitigate the impact of aging on the fuel cell system performance, good corrective actions must be taken. To do so, the performance prediction during the aging of the fuel cell system must be conducted. In this paper a verified and tested prognosis approach applicable to fuel cells is presented. The novelty in the approach used is linked to its modular structure. In fact, this approach has three phases. The first one aims to identify the internal physical parameters of the stack using an optimization algorithm on experimental data and a static fuel cell model. The second one predicts the temporal evolution of these parameters using a prediction algorithm. Finally, the reconstruction phase consists in the prediction of parameters that are re-injected into the model to reconstruct polarization curves and thus reflect the performance degradation of the fuel cell. The reliability and repeatability of the proposed approach have been successfully validated on two data sets from two different experimental campaigns.