Body fat stores may serve as an index of condition in mammals. Thus, techniques that measure fat content accurately are important for assessing the ecological correlates of condition in mammal populations. We compared the ability of two conductive techniques, bioelectrical impedance analysis (BIA) and total body electrical conductivity (TOBEC), to predict body composition with that of morphometric methods in three small mammal species: red squirrels (n=13), snowshoe hares (n=30), and yellow-bellied marmots (n=4). Animals were livetrapped in northern Idaho; BIA (all subjects) and TOBEC (squirrels only) measurements were taken following chemical immobilization in the field, and morphometric measurements were taken postmortem. Information provided by BIA and TOBEC failed to improve upon the predictive power of morphometric equations for total body water (TBW) and lean body mass (LBM) in squirrels and hares, which do not store substantial amounts of fat (<5% body mass comprised of fat). Although the same pattern held with respect to LBM in marmots, which accumulate substantial amounts of body fat (>10% body mass), a BIA-based model proved best at estimating TBW, suggesting that the usefulness of conductive techniques may be a function of fat deposition. However, regardless of the technique used to predict body composition, estimates of body fat furnished by our equations failed to approximate actual fat levels accurately in all three test species, probably because these techniques only provide indirect estimates of fat content. These results highlight the limitations inherent in contemporary methods of animal fat estimation and underscore the need for the development of direct and accurate measures of body fat in mammals.