The importance of verifying the calibrating numerical models, before using them to predict design criteria, is well known. A verification scheme for wave refraction models has recently been developed. Two of the latest wave refraction models have been acquired and tested The first is based on a parabolic approximation method, using a finite difference technique, which includes the effects of combined refraction/diffraction, wavecurrent interaction, partial and full wave breaking and non-linear effects The second is a simpler model based on the wave action equations, and includes the effects of wave refraction, shoaling, partial breaking and bottom friction. The first model is most useful over coastal areas (up to 200 km2) but is relatively expensive to run compared with the wave action model. The second model, although less rigorous in its solution of the physical equations, is more useful over much larger areas (up to 2000 km2). The method has been carefully designed to test all aspects of model performance. This includes: (1) Verification against laboratory data (where traditional ray techniques break down) (2) Sensitivity tests against hypothetical bathymetnes for several input/boundary conditions. (3) Verification against two field data sets, both including the measurement of directional spectra This chapter outlines the test procedure and presents the results using the two models described above. It also highlights areas where development work on the models was necessary, and where future work is required. Development of the modelling techniques is continuing.