Scatterometry is a common technique for the characterization of nano-structured surfaces. The goal is to establish scatterometry as a traceable and absolute metrological method for dimensional measurements. Scatterometry is an indirect measuring method. There are three aspects to be investigated for a proper estimation of the measurement uncertainties. Firstly, the measured optical properties will have measurement uncertainties. The optimization theory provides well-proven algorithms to propagate the uncertainties of the input parameters to the output, using covariance or Monte Carlo methods. Secondly, it must be assured that the model represents the measurement object sufficiently well. Due to the large number of parameters it takes to model a real object, it is not possible to simultaneously determine all of them from the measured data. Therefore, a priori knowledge about reasonable values for certain model parameters and their associated variances is needed. Their impact on the uncertainties of the reconstructed parameters describing the nano-structured surface has to be considered too. Thirdly, there are assumptions in the model which are inherent properties of the mathematical methods used. The assumption that the structures are periodic over a sufficiently large area, i.e. the measurement field, is of the most importance. Line-edge or line-width roughness are obviously effects disturbing the periodicity and the corresponding uncertainty contribution can be derived by an analytical approach to estimate their influence on the diffracted intensities. In our example the latter contribution is the dominant one.© (2011) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.