Discrete element method (DEM) computer simulations are used to investigate intra-tablet coating thickness variability. Two new post-processing algorithms are presented. The first algorithm uses an image-based method to track the exposure to a simulated spray of small area panels on each tablet's surface so that the distribution of spray exposure times over the tablet's surface can be determined directly from DEM data. The second algorithm predicts the asymptotic limit of intra-tablet coating uniformity. This second algorithm includes the influence of tablet orientation and shadowing when considering exposure to the spray, averaged over many tablets. The DEM simulations produce the first direct evidence that non-spherical tablets approach asymptotic intra-tablet coating variability values. The asymptotic limits are predicted well using the new asymptotic prediction model. In general, tablet caps have thicker coatings than tablet bands. Moreover, tablets that have a more elongated shape tend to have less coating on the smaller radius of curvature portions of the bands. Of particular importance in this new asymptotic modeling approach is the inclusion of shadowing effects. When shadowing is not included and only tablet orientation is considered, the predictions over-predict the asymptotic intra-tablet coating variability values and also change the observed rank order of the asymptotic values for different tablet shapes. The asymptotic intra-tablet coating variability values using the new algorithm correlate reasonably well with tablet sphericity, with increasing sphericity improving coating uniformity. This paper also presents the first attempt to directly compare experimental and simulated coating thickness distributions. The asymptotic coating thickness predictions compare well qualitatively with terahertz thickness measurements made on tablets from coating experiments. Unfortunately, only qualitative comparisons could be made due to the limited number of tablets sampled experimentally and differences in spray zone areas and flux distributions. The tablets in the experiments, however, displayed similar features as those found in the simulations. (C) 2015 The Authors. Published by Elsevier Ltd.