Comparisons made between RCS calculations and measurements for different types of combat vehicles at 35 GHz show differences which can be explained by that the surfaces of the cad-models used in the calculations were too smooth. I real case a surface can show stochastic structure which lead to a reduction of the RCS levels in specular direction while, increase in side lobe directions. In SPECRAY EM there are tools, that by surface texturation, takes surface undulations into account by so called bump mapping. In this work the bump mapping algorithm has been validated against IEM. The results show very good agreement for the coherent scattering, which in many cases are the dominant part of the scattering, for all kinds of surfaces. A very good agreement is seen for the non-coherent part when the surface average slope is small and the incident wavelength is small compared to the correlation length of the surface structure. If the wavelength is long compared to the surface structure the results are acceptable. Bump-mapping tend to overestimate HH-polarisation contribution whereas the VV-polarisation contribution is underestimated. When the internal surface slopes is large, multiple scattering terms not include in the bump mapping algorithm, will be significant, hence the non-coherent scattering will be underestimated. The overall conclusion is that bump mapping is competent tool used in RCS-calculations.