A case study of the development of an overall resource assessment for the Wave Hub site in the southwest UK is presented. Wave Hub is one of the earliest large-scale wave farms planned. Several resource assessment studies have been performed for the site, but the published results are high-level and predicted power availability varies significantly. This paper provides a detailed overview and re-analysis of the multiple datasets used in the original studies, which consisted of a combination of physical measurement and numerical modelling. The quality of the datasets is assessed, and reasons for the discrepancies between predicted resource levels investigated. Results from a SWAN model for the region illustrate significant levels of spatial variability in the resource due to the complexity of the local bathymetry, and examination of long-term global model datasets shows notable inter-annual variability. It is thus concluded that a resource assessment methodology utilising datasets from multiple locations and of short duration significantly reduces the accuracy of the predicted levels of resource. From these results, key learnings for future developments are discussed. (C) 2013 Elsevier Ltd. All rights reserved.