The observation that "on the whole... larger species live farther north and the smaller ones farther south" was first published by Carl Bergmann in 1847. However, why animal body mass might show such spatial variation, and indeed whether it is a general feature of animal assemblages, is currently unclear. We discuss reasons for this uncertainty, and use Our conclusions to direct an analysis of Bergmann's rule in the mammals in northern North America, in the communities of species occupying areas that were covered by ice at the last glacial maximum. First, we test for the existence of Bergmann's rule in this assemblage, and investigate whether small- and large-bodied species show different spatial patterns of body size variation. We then attempt to explain the spatial variation in terms of environmental variation, and evaluate the adequacy of our analyses to account for the spatial pattern using the residuals arising from our environmental models. Finally, we use the results of these models to test predictions of different hypotheses proposed to account for Bergmann's rule. Bergmann's rule is strongly supported. Both small- and large-bodied species exhibit the rule. Our environmental models account for most of the spatial variation in mean, minimum and maximum body mass in this assemblage. Our results falsify predictions of hypotheses relating to migration ability and random colonisation and diversification, but support predictions of hypotheses relating to both heat conservation and starvation resistance.