Although progress has been forthcoming concerning the spatial scaling of macroecological patterns, less attention has been focused on their temporal dynamics. The geohistorical record of Quaternary mammals offers an opportunity to examine macroecological patterns on longer temporal scales. Recent work has shown that the shapes of mammalian body size distributions vary with latitude. Temperate communities show relatively uniform distributions regardless of continent. Moreover, over the last 40 ky mammalian body size distributions of temperate North America are relatively flat despite considerable turnover in community composition and an extinction event of large bodied mammals. Recent tropical communities show a wide range of variation in the shapes of body size distribution resulting in a latitudinal gradient in the mean, median and kurtosis. Current work evaluates the relationship between environmental variables and the shapes of mammalian body size distributions. The mean, median, skew and kurtosis for each Recent local assemblage was calculated and examined for patterns with respect to mean annual temperature and rainfall as well as mean January and July temperature and rainfall. The relationship between the shapes of the mammalian body size distributions and climate was weak, but significant on all continents. However, analyses of the shapes of the distributions with respect to habitat type indicate that the productivity and dimensionality of the habitat is a better predictor of the shapes of the distributions. For example, highly productive tropical forests tend to have more peaked distributions than grassland or desert communities. Repeating this analysis with late Pleistocene mammals communities for which we know the habitat type indicates that this relationship is consistent over the last 40 ky. The similarity between the body size distributions of Late Quaternary assemblages and Recent assemblages implies that body size plays a consistent role in local community assembly. Moreover that role is mediated by environmental variables.