Mesowear analysis is a form of dental wear analysis used to infer the diets of herbivorous mammal species. It makes use of percentage indices of blunt, round and sharp cusp shape and high occlusal relief to classify the diet of species into one of three categories: browser, grazer or mixed feeder. Previously, this form of analysis has been limited to placental mammals, restricting the use of such analyses in Australia where the dominant herbivorous mammalian fauna consist of marsupials. In order to address this limitation, mesowear variables of extant marsupials were examined to determine whether their diets can accurately be predicted using mesowear analyses. Discriminant Function Analysis of mesowear variables and analysis of variance (ANOVA) of univariate mesowear scores for marsupial species demonstrate that mesowear analysis can be used to classify marsupial diets. A dataset of 24 typical marsupial species considered to be representative of the three dietary categories with respect to mesowear was generated and significantly increased cross-validated classification levels from 74.4% to 100% for the second molar. Mesowear analysis for marsupial species is most effective for the second molar with high predictive power also being evident for the first and third molars. When mesowear analysis was applied to six Plio-Pleistocene macropods (Marsupialia: Macropodidae) from the Darling Downs region, southeast Queensland, all species were classified as mixed feeders with the exception of Protemnodon roechus which was classified as a grazer. This study demonstrates the effectiveness of mesowear analysis as a dietary proxy for herbivorous marsupial species.