1. Estimates of body condition are regularly made in wildlife studies, particularly those focused on individual and/or population performance; however, many studies assume that it is always beneficial to be heavier or have a higher body condition index (BCI), without accounting for the physiological significance of variation in the composition of tissues that differ in their function such as fat and lean mass. 2. We hypothesized that the relationship between BCI and masses of physiologically important tissues (fat and lean) would be conditional on the annual patterns of energy acquisition and expenditure of individuals under study, and tested relationships in three species with contrasting ecologies in their respective natural ranges: an obligate hibernator (Columbian ground squirrel,Urocitellus columbianus), a facultative hibernator (black-tailed prairie dog,Cynomys ludovicianus), and a food-caching non-hibernator (North American red squirrel,Tamiasciurus hudsonicus). 3. We measured fat and lean mass in adult males and females of these three species using quantitative magnetic resonance (QMR). We measured body mass, two measures of skeletal structure (zygomatic width and right hind foot length) to develop sex- and species-specific BCIs, and tested the utility of BCI to predict body composition in each species. 4. Body condition indices were more consistently and more strongly correlated with lean mass than fat mass. The indices were most positively correlated with fat when fat was expected to be very high (pre-hibernation prairie dogs). However, in all cases, fat and lean mass were better predicted by overall body mass rather than BCI.