Degradation issues remain among the main factors impeding the development of commercially viable fuel cell (FC) systems for transportation and stationary power applications. The necessity to keep costs and performances at reasonable levels led to the widespread use of carbon supported platinum (Pt) for electrodes and thin PFSA membranes as electrolyte. Both are quite fragile materials but currently without any convincing alternative. To better understand, and eventually limit degradation mechanisms, several kinds of accelerated stress tests (AST) have been developed targeting each material independently: namely open circuit voltage (OCV) for PFSA membranes, potentiostatic regimes for carbon, and potential cycling for Pt [1, 2]. Other accelerated stress tests corresponding to common fuel cell operation sequences were also proposed like constant current, start/stop, hydrogen starvation/dead-ended anode protocols [3,4], temperature, relative humidity or potential/load cycling, or even freeze/thaw protocols [5]. In this work, we focus on load cycling events, which are very close to the potential cycling performed during AST to study Pt and carbon degradation mechanisms. For obvious reasons, load cycling in actual fuel cell systems and potential cycling in AST are usually performed in overfed conditions [6]: i.e. the fuel cell is supplied with the gas flow rates corresponding to the highest current before the intensity increases. And in most of the cases, the gas flow rates are even permanently kept at the highest values during AST. To avoid supplying fuel and air in excess, most of fuel cell stacks are hybridized either with batteries or supercapacitors (SC) to improve the system dynamics (which in return increases its size and cost). Thanks to hybridization, the FC voltage can drop by a few hundreds of mV per cell during the time needed by the gas supply lines to adapt the flow rates, while batteries or SC keep the system power at the required value. However, the impact of such short reactant starvation events on FC performance and durability is not very well known. Therefore, we performed repeated current Heaviside steps as AST on PEM single cells; typical values were from 0.25 to 0.95 A/cm² (with corresponding FC voltages between 0.65 and 0.8 V). The gas flow rates were set either to the maximum current intensity (with air and H2 stoichiometries of 3 and 1.5 in open-anode or 1 in dead-end mode, respectively) or we let the gas lines adjust the flow rates when the current input changed, which took about half a second (during this time, the FC could fell to about 0.1 V). As shown on the Figure, the performance and electrochemical surface area repeatedly aged significantly slower in the second case, i.e. with short gas starvation events. To better understand this difference, similar experiments were performed in a segmented cell [7] with measurements of local currents, anode and cathode potentials. We observed that local cathode potentials were higher in overfed conditions while they dropped for a short time in the other case due to (most probably) oxygen depletion in the cathode compartment. Local negative currents were also observed sometimes when the load decreased, but since they were not associated to high local potentials, they did not worsen degradation. In the same way, since the FC dynamics were limited most probably by air supply, the anode (and thus cathode) local potentials remained low. These results suggest that FC voltage drops during transients due to short lacks of reactants can improve the durability of the catalyst layer, as well as optimize the system efficiency in terms of hydrogen consumption and air compressor power demand which can be a useful observation for the design of optimized hybridization strategies.