This paper addresses the common uncertainties introduced into the long-term performance estimation of wave energy converters in coastal regions by using 3 rd -generation model wave data as input of the simulator. Indeed, the variability observed in hindcast results induces variability in the energy converted by devices, with implications on the long-term power figures. A comparison with estimates based on three different wave model data sources (produced from the WAM, Wave Watch III and SWAN wind-wave models) and buoy measurement data in the nearshore region of the French marine energy test site SEM-REV is made. The paper raises the fact that, in general, a non-negligible error is obtained – even when the wind-wave model has been appropriately designed to the local wave physics – with respect to real converted power values, which in turn results in significantly biased long-term averaged power numbers.