DNA methylation is an important, heritable epigenetic modification in most eukaryotic organisms that is connected with numerous biological processes. To study the impact of natural epigenetic variation in an ecological or evolutionary context, epigenetic studies are increasingly using methylation-sensitive amplification polymorphism (MSAP) for surveys at the population or species level. However, no consensus exists on how to interpret and score the multistate information obtained from the MSAP banding patterns. Here, we review the previously used scoring approaches for population epigenetic studies and develop new alternatives. To assess effects of the different approaches on parameters of epigenetic diversity and differentiation, we applied eight scoring schemes to a case study of three populations of the plant species Viola elatior. For a total number of 168 detected polymorphic MSAP fragments, the number of ultimately scored polymorphic epiloci ranged between 78 and 286 depending on the particular scoring scheme. Both, estimates of epigenetic diversity and differentiation varied strongly between scoring approaches. However, linear regression and PCoA revealed qualitatively similar patterns, suggesting that the scoring approaches are largely consistent. For single-locus analyses of MSAP data, for example the search for loci under selection, we advocate a new scoring approach that separately takes into account different methylation types and thus seems appropriate for drawing more detailed conclusions in ecological or evolutionary contexts. An R script (MSAP_score.r) for scoring and basic data analysis is provided.