Regional assessments of the wave energy resource tend to focus on averaged quantities, and so provide potential developers with no sense of temporal variability beyond seasonal means. In particular, such assessments give no indication of inter-annual variability – something that is critical for determining the potential of a region for wave energy convertor (WEC) technology. Here, we apply the third-generation wave model SWAN (Simulating Waves Nearshore) at high resolution to assess the wave resource of the northwest European shelf seas, an area where many wave energy test sites exist, and where many wave energy projects are under development. The model is applied to 7 years of wind forcing (2005–2011), a time period which witnessed considerable extremes in the variability of the wind (and hence wave) climate, as evidenced by the variability of the North Atlantic Oscillation (NAO). Our simulations demonstrate that there is much greater uncertainty in the NW European shelf wave resource during October–March, in contrast to the period April–September. In the more energetic regions of the NW European shelf seas, e.g. to the northwest of Scotland, the uncertainty was considerably greater. The winter NW European shelf wave power resource correlated well with the NAO. Therefore, provided trends in the NAO can be identified over the coming decades, it may be possible to estimate how the European wave resource will similarly vary over this time period. Finally, the magnitude of wave power estimated by this study is around 10% lower than a resource which is used extensively by the wave energy sector – the Atlas of UK Marine Renewable Energy Resources. Although this can partly be explained by different time periods analysed for each study, our application of a third-generation wave model at high spatial and spectral resolution significantly improves the representation of the physical processes, particularly the non-linear wave-wave interactions.