At the most fundamental level, the size of an animal's home range is determined by its energy needs. In the absence of confounding variables, home range size should therefore scale with body mass according to Kleiber's exponent for metabolic rate of 0.75. Comparative studies in a wide range of taxa have failed to confirm this prediction: home range size has commonly been found to scale with an exponent significantly >0.75. We develop a comparative measure of metabolic needs that incorporates both mass-specific metabolic rate and social-group size. We test the prediction that home range size in primates scales isometrically with this measure when an appropriate linear model is applied to data corrected for phylogenetic bias. Analyses using species values as data points indicate an exponent consistent with Kleiber's law. This result is misleading, however, because ecological factors confound the analysis, and the slopes within some ecologically homogeneous taxa are steeper. Accordingly, in analyses based on independent contrasts with reduced major axis, slopes are significantly greater than predicted by Kleiber's law. We examine the effects of other variables, and we find that systematic variation in substrate use, home range overlap, and diet account for the steeper than expected relationship between home range size and metabolic needs based on Kleiber's law. We therefore conclude that the scaling of home range size is subject to Kleiber's law but in combination with other factors. These results emphasize that the study of allometry requires detailed attention to statistical models and control of confounding variables.