Surface/interface roughness has a significant influence on the shape of the depth profile measured by any depth profiling technique. Such an influence is particularly significant for thin delta layers and at sharp interfaces of single-and multilayers. In the mixing-roughness-information (MRI) model for quantification of measured depth profiles, the influence of roughness is usually taken into account by a Gaussian height distribution function (HDF). If the roughness cannot be represented by a Gaussian HDF, a non-Gaussian HDF has to be implemented into the MRI model. Deviations of simulated depth profiles using the MRI model with Gaussian and with several well-defined non-Gaussian HDFs are evaluated quantitatively. The results indicate that a realistic non-Gaussian HDF has to be taken into account if high accuracy in quantification of sputter depth profiles is required. Of particular importance is the case of a roughness given by an asymmetrical HDF. Application of an asymmetrical triangle height distribution function in the MRI model yields an excellent fit for the measured AES depth profiling data of a polycrystalline Al film. (C) 2013 Elsevier B.V. All rights reserved.