What determines the lengths and proportions of mammalian limbs? While the answer to this question is still largely unknown, a number of workers have recently begun analysing the selection of morphology in a rigorous framework, searching for quantitative links between structure, performance, and their ecological and behavioural context. The present study investigates a variety of ecological and behavioural variables to determine whether or not they are correlated with hind-limb length in the Carnivora. Data were analysed by using phylogenetically independent contrasts and phylogenetic analysis of covariance. We found that traditional perceptions of limb length are often inaccurate; some species widely regarded as relatively long-legged actually have limb lengths near those expected for their body size. Interestingly, relative hind-limb length is not a significant predictor of distance moved daily, home-range area, or prey size. Phylogenetic ANCOVA results, however, indicate a relationship between prey-capture behaviour and relative hind-limb length. These findings suggest that the evolution of carnivoran limb length has been most influenced by selection for prey-capture behaviour. These results, coupled with those of other studies, can be used to suggest which performance variables could most fruitfully be studied in the laboratory to understand the selection of the structure of the mammalian limb. We suggest that relevant performance variables might be: maximum jump height and/or length, the ability to generate outforces, and levels of stress tolerance in limb bones during prey pursuit/capture.