For the first time a prediction model of regular morphological patterns on the seabed was tested against observations of sand wave and sand bank occurrence in the entire North Sea. The model, which originates from first physical principles, predicts this occurrence via two dimensionless parameters on the basis of the water depth, the tidal velocity amplitude, the level of zero intercept above the seabed (z0), and a viscosity variation parameter ϵ. The latter two quantities were varied in a number of predictions for the entire North Sea, whereas for the first two, local values were used. The range of realistic values of ϵ and z0 was large enough to let these two parameters distinguish between the possible (combinations of) bed forms, as is shown in the shallower southern bight of the North Sea. The results were more sensitive to variations in z0 than in ϵ. A slightly more detailed approach focused on sand waves only in the southern North Sea and estimated local values for z0 using depth information. Quantification of the results showed that the model was able to predict the contours of the sand wave patches, but it could not account for the absence of the bed features within this area. The type of bed deposit partly explains the smaller-scale variation. The work confirms the validity of the theoretical bed form prediction model and verifies the hypothesis that the large-scale seabed features are formed as free instabilities of tide-topography interactions. ¿ 2001 American Geophysical Union