Recent years have seen an increased emphasis on measuring ecologically relevant performance capabilities to undersand associations between morphology and habitat use.  Such studies presume that performance is invariant, but in eight Caribbean Anolis lizard species, we found that maximum sprinting ability depends on surface diameter.  Moreover, these species differ in the degree to which sprint speed declines with decreasing surface diameter, defined as "sprint sensitivity" (high sprint sensitivity = substantial declines in speed between broad and narrow dowels).  The habitat constraint hypothesis postulates that Anolis lizards will avoid structural habitats in which their maximal sprinting capabilities are impaired.  The habitat breadth hypothesis postulates that species whose performance is less affected by substrate will use a greater variety of habitats than species whose performance varies to a greater extent on surfaces of different diameters.  Field observations quantified the proportion of time that lizards spent on different perch diameters.  Both hypotheses were confirmed: species with high values of sprint sensitivity avoided using perches on which their maximal sprinting abilities are impaired, whereas species with low sprint sensitivity used such "submaximal" surfaces more frequently.  Species with low sprint sensitivity used a broader range of structural habitats than species with high sprint sensitivity.