AimMitigating the effects of extreme conditions is a mechanism that can structure the activity patterns and habitat selection of a species and may particularly impact species at the extremes of their geographic distribution. Furthermore, changing climate patterns have the potential to influence biotic interactions between species in novel ways. As two species at the edges of their northern and southern distributions, respectively, ocelots and bobcats may face unique pressure compared to individuals in more central portions of their range. Our objective was to describe the selection of thermal cover by ocelots and bobcats to examine whether partitioning of thermal resources was occurring or if this was a source of potential interspecific competition.LocationWe monitored eight ocelots and 13 bobcats in South Texas, USA.MethodsWe compared selection within, above, and below their estimated thermoneutral zones to examine the effect of varying temperature on habitat selection. Additionally, we stationed 130 black globe thermal sensors to describe the thermal properties of the various cover types.ResultsWe observed variation in habitat selection across temperatures and species. Ocelots and bobcats selected for shrub cover and vertical canopy cover when cold stressed. When heat stressed, both species avoided bare ground and selected for higher vertical canopy cover and were located closer to dense cover. Black globe measurements revealed differences in environmental temperature across cover types, with forest and shrub cover significantly lower than herbaceous or bare ground.Main conclusionsChanging climates may influence interspecific competition and alter areas of sympatry through range shifts. Our results stress the importance of dense shrub cover and forested canopy as thermal refuge for ocelots and bobcats and suggest that abundant vegetation may mitigate the effects of interspecific competition during lower temperatures and niche partitioning may reduce interspecific competition during upper temperature limits, providing support for the interactive range-limit theory.