There is a long tradition of neotaphonomic research in paleontology and archaeology that is intended for building models of the bone-accumulating and -modifying behaviors of prehistoric biological actors. Building on that tradition, we present new data comparing the ways by which wild and captive leopards (Panthera pardus) altered the meat-bearing long limb bones of small bovids on which they fed. Compared to wild leopards, a captive leopard created more than twice as many tooth marks per skeletal element. In addition, the captive leopard imparted much more intensive gross damage to bones than did wild leopards—including complete consumption of some bones or deletion of bone portions (especially epiphyses) and extensive fragmentation of diaphyses. We attribute the stark divergence of taphonomic results in the bone samples formed by wild and captive leopards to environmental variability: the wild samples were modified by leopards operating in systems characterized by low to moderate feeding competition; the captive sample was created by a leopard operating in the complete absence of feeding competition but (unlike the wild leopards) was also subjected to the tedium of confinement—a condition that prompted repeated and prolonged bouts of chewing beyond what was required to simply extract nutrients from bones. The variable condition of the two bone samples also represents opposing endpoints of leopard taphonomic potential—the wild sample reflecting minimal expected alteration and the captive sample reflecting maximal expected alteration. Neither endpoint serves as accurate proxy for expectations of the taphonomic alteration of small bovid bones by leopards operating in a natural setting that is characterized by moderate to high feeding competition, such as Tanzania's modern Serengeti ecosystem or that reconstructed for most African Pliocene and Pleistocene habitats. Our results do, however, reinforce the conclusions of a previous study comparing taphonomic patterns created by wild and captive lions (P. leo)—that environmental context is a key variable to control if the analyst's goal is to produce the most scientifically realistic and usefully applicable models of the bone-accumulating and -modifying behaviors of prehistoric biological actors.