The bed of coastal seas displays a large number of rhythmic bed features, of which sand waves are relevant to study from an engineering perspective. Sediments tend to be well‐sorted over these bed forms, which is so far poorly understood in terms of modeling of finite‐amplitude sand waves. Using Delft3D, we employ bed stratigraphy and consider four different grain size classes, which are normally distributed on the phi‐scale. The standard deviation (sortedness) is then varied, whereas the geometric mean grain size is kept constant. The results show that typically the crests of sand waves are coarser than the troughs. Residual flow causes net sedimentation on the lee side of the crest and, consequently, the general sorting pattern is distorted. Since larger grains experience a larger settling velocity they are deposited on the upper lee slope, whereas the smaller grains are found on the lower lee slope. Due to sand wave migration, also the internal structure of the sand wave is revealed, which follows the same sedimentation pattern as the lee slope surface. These results qualitatively agree with sorting patterns observed offshore. The sorting processes lead to longer wavelengths and lower wave heights, as a function of standard deviation. This relates to the dampening effect of suspended sediment transport for fine grains. Finally, it appears that the modeled wave heights fall in the same range as observations in the North Sea. These results are valuable for e.g. predicting the morphological response after engineering activities and determining suitable aggregates for sand extraction.