The basal rates of metabolism (BMR) of bats belonging to the family Phyllostomidae are re-examined after an earlier correlation with food habits was rejected because it did not take phylogeny into consideration. This rejection was based on an erroneous attribution of food habits and on an analytical method, phylogenetic contrasts, that ignores interactions that occur among character states and preferentially attributes responsibility for character states to phylogeny. The re-examination made here was based on analysis of covariance, which makes no a priori assumptions on the relative impact of factors that influence character states and permits factor interactions to be identified. A resulting model, based on variation in body mass, food habits, occurrence with respect to elevation, and residence on islands or continents, accounts for 99.4% of the variation in the BMR of 30 species of phyllostomids. Basal rate is also correlated with subfamily, but only if food habits are excluded because they are correlated with subfamily affiliation, as is residence on islands and continents, two examples of factor interaction. The preference to assign the effects of food habits and island residence on basal rate to subfamily affiliation (and phylogeny) is not justified. The concept that quantitative physiological characters can be transmitted via phylogeny without regard to the habits of animals and the characteristics of their environments cannot be defended. Phylogeny is the historical context in which the evolution of character states occurs, not the 'cause' of their evolution.