Summary1. Aspects of locomotor performance are thought to be important in a variety of natural contexts, including foraging and escaping from predators. The role of locomotor abilities in sexual selection is less well documented, but is probably of wide importance. 2. Sexual selection may exaggerate traits beyond their optimum with respect to biomechanical or physiological function, thus imposing 'costs' of various types. Locomotor performance, in conjunction with its lower-level determinants and its behavioural and ecological consequences, constitutes an ideal set of phenotypes with which to study such costs. 3. Organisms may possess traits that compensate for the negative effects of sexually selected traits on performance abilities. Ignoring these may lead to erroneous conclusions about the cost of a putatively sexually selected trait. At the same time, ignoring the effects of sexual selection on functionally important traits may lead us to underestimate their functional significance. 4. Many properties of organisms are affected by, or at least correlate with, overall body size, so this must be considered in any study of adaptation or sexual selection. For interspecific or interpopulation comparative studies, phylogeny must also be considered in statistical analyses. 5. We present a hypothetical data set that illustrates the perils of ignoring potential compensatory mechanisms when studying the costs of a sexually selected trait. We also discuss different ways of studying costs, including analyses of a single sex and of the differences between the sexes.