PACS 68.35. Ct, 82.80.Pv Angle-resolved X-ray photoelectron spectroscopy (ARXPS) is a method for non-destructive depth profiling in the nm thickness range. For the quantification of the data, usually model calculations are used which approximate the surface structures by smooth layers. Using ARXPS data from a computer simulation, in this paper we study the influence of overlayer roughness on the quantification results. It can be followed, that rough overlayers modeled by means of small-sized islands are identified nearly as smooth layers. The effective thickness, however, can be evaluated always satisfactorily.1 Introduction The current trend to micro-scaled materials immediately leads to the demand of a highsophisticated analysis of such nm-sized features. Besides an enormous development in transmission electron microscopy, also the classical surface spectroscopic analytical methods there are still in the focus of interest. One promising method for non-destructive depth profile analysis is the angle-resolved X-ray photoelectron spectroscopy (ARXPS). The basis for the low depth information range at XPS of only some nm is the attenuation of the used signal electrons by inelastic scattering. For ARXPS the effect is used that the effective information depth can be changed by variation of the electrons take-off angle during measurement. In series of XPS measurements with angle variation, depth information about the first some nm is found, however, for quantitative interpretation always model calculations are necessary. During this process the ARXPS measurement results were compared with those from intensities calculated from an assumed model structure of the surface region. The parameters (thicknesses, stoichiometry, surface coverage) of the model structures are varied and the best approximation for calculated to experimental results is searched for. This is a complicated process because the results found are very sensitive to the model assumptions and boundary conditions used. Already a couple of years ago Cumpson [1] showed theoretically that the number of parameters that can be derived from ARXPS measurements is very limited.Because all of this, mostly simple models with smooth layers (at least with islands on top) [2-4] are used for the ARXPS quantification. However, it is quite clear that this is an incorrect assumption in many of the real cases [5]: all realistic surfaces have a roughness and may have lateral inhomogeneities. Thus it can be clearly assumed that such effects can influence the found results. In order to quantify such possible misinterpretation we decided to make computer simulations of ARXPS results for more complicated surface structures. In a recent work [6] we proposed an algorithm and showed, beside some remarks to the "magic angle concept" at overlayers on rough surfaces, in particular first results for island structures on top of smooth surfaces. It was concluded there that an overestimation of the islands surface coverage is a typical artifact when the dimensions of the islands are in t...