1. Life-history theory predicts the progressive dwarfing of animal populations that are subjected to chronic mortality stress but the evolutionary impact of harvesting terrestrial herbivores has seldom been tested. In Australia, marsupials of the genus Macropus (kangaroos and wallabies) are subjected to size-selective commercial harvesting. Mathematical modelling suggests that harvest quotas (ca. 10–20 % of population estimates annually) could be driving body-size evolution in these species. 2. We tested this hypothesis for three harvested macropod species with continental-scale distributions. To do so, we measured more than 2 000 macropod skulls sourced from wildlife collections spanning the last 130 years. We analysed these data using spatial Bayesian models that controlled for the age and sex of specimens as well as environmental drivers and island effects. 3. We found no evidence for the hypothesized decline in body size for any species; rather, models that fit trend terms supported minor body size increases over time. This apparently counterintuitive result is consistent with reduced mortality due to a depauperate predator guild and increased primary productivity of grassland vegetation following European settlement in Australia. 4. Spatial patterns in macropod body size supported the heat dissipation limit and productivity hypotheses proposed to explain geographic body-size variation (i.e., skull size increased with decreasing summer maximum temperature and increasing rainfall, respectively). 5. There is no empirical evidence that size-selective harvesting has driven the evolution of smaller body size in Australian macropods. Bayesian models are appropriate for investigating the long-term impact of human harvesting because they can impute missing data, fit non-linear growth models and account for non-random spatial sampling inherent in wildlife collections.