The new EGM2008 global gravity field model is evaluated by comparisons of geoid heights computed from the model with those available at GPS levelling stations in various regions and by computing sea surface topography solutions from the difference between the mean sea surface and the geoid from this model. In order to identify how good the model performs the same tests also are performed for other recent global gravity field models. The evaluation method, in particular, has to take into account the omission error when truncating global gravity field models at chosen degrees and orders and when comparing them with observed quantities. The procedures applied for the computation of the omission error as well as the general methods applied for the evaluation are described in the paper. For testing the models there are available GPS levelling heights in six different regions (Europe, Germany, USA, Japan, Canada, Australia). Some of these data sets seem not to be adequate for evaluation purposes, because they exhibit long wavelength structures in the geoid height differences. Nevertheless, from the results of the geoid height comparisons one can conclude that the EGM2008 model performs best in most of the regions. Similar results are derived from the sea surface topography solutions based on different global gravity field models. The dynamic ocean topography determined with the EGM2008 geoid seem to show more realistic and more detailed features than solutions based on other global gravity field models. In summary, from the tests performed one can state that EGM2008 represents a significant improvement of the global gravity field in terms of quality and resolution.