Extreme value analysis is used to assess the impact of the new approaches described in the LOADS JIP on the development of metocean wave criteria in the Central North Sea. The use of the Forristall distributions for determining crest and wave heights is well established but this paper compares inferences using those short-term probability distributions with those from the new approaches in LOADS which better reflect wave non-linearity and the effects of wave breaking in intermediate and deep waters. The new distributions also indicate some sensitivity to wave spreading and JONSWAP peakedness and the sensitivity of extreme wave conditions to these sea-state descriptors is examined. For crest heights, the analysis covered both point and area statistics for nominal deck sizes up to 50m x 50m. In the presence of uncertainty, there is also potential ambiguity surrounding the precise definition of return value. The quantitative effect of uncertainty is dependent upon the definition used and the impact of three different definitions is compared with the traditional approach of effectively ignoring uncertainty in the extrapolation process. The analysis was carried out using the NORA10 WAM hindcast developed by the Norwegian Met Office which covered approximately 60 years of historical atmospheric and sea-state conditions. The paper presents quantitative comparisons between return values derived using the different approaches concentrating on return periods of 100 and 10,000 years.