In the field of biomaterials surfaces, the ability of the atomic force microscope (AFM) to access the surface structure at unprecedented spatial (vertical and lateral) resolution, is helping in a better understanding on how topography affects the overall interaction of biological cells with the material surface. Since cells in a wide range of sizes are in contact with the biomaterial surface, a quantification of the surface structure in such a wide range of dimensional scales is needed. With the advent of the AFM, this can be routinely done in the lab. In this work, we show that even when it is clear that such a scale-dependent study is needed, AFM maps of the biomaterial surface taken at different scanning lengths are not completely consistent when they are taken at the same scanning resolution, as it is usually done: AFM images of different scanning areas have different point-to-point physical distances. We show that this effect influences the quantification of the average (R_a) and rms (R_q) roughness parameters determined at different length scales. This is the first time this inconsistency is reported and should be taken into account when roughness is measured in this way. Since differences will be in general in the range of nanometres, this is especially interesting for those processes involving the interaction of the biomaterial surface with small biocolloids as bacteria, while this effect should not represent any problems for larger animal cells.