Large carnivores commonly scavenge on kills made by other species, but if and how this phenomenon is influenced by urbanization remains unclear. To address this knowledge deficit, we investigated whether housing density, along with demographic and environmental covariates, impacted the probability of American black bear (Ursus americanus) occurrence at cougar (Puma concolor) killed prey along the wildland-urban gradient of western Washington, USA. Under the refuge hypothesis, which stipulates that residential development reduces opportunities for black bears to visit cougar prey carcasses by (1) altering cougar kill composition and/or (2) drawing black bears to human subsidies, we expected the probability of bear presence at cougar kills to decline as housing density increased. Alternatively, under the pileup hypothesis whereby reduced green space drives a greater overlap and thus more frequent interactions among carnivores, we predicted that bear presence at cougar kills would increase with housing density. Occupancy models derived from forensic and remote camera evidence of bear visitation to carcasses at kill sites identified from 12 GPS-collared cougars indicated that the probability of bear presence at kill sites decreased when cougars foraged on small-bodied prey, increased in summer compared with autumn, and declined with increasing housing density. Indeed, the top model indicated a multiplicative decrease of 500 in the odds of black bear carcass visitation for every additional house per hectare on the landscape, supporting the refuge hypothesis. These results suggest that residential development has the potential to alter intraguild relationships among large carnivores, even at modest levels where robust carnivore populations persist on the landscape, and may alter scavenger dynamics at carcasses where black bear presence is virtually eliminated.