Although diversionary feeding has been used as a management strategy to limit problem bear behaviors or reduce human–wildlife conflict, the underlying mechanisms determining the success or failure of diversionary feeding programs are infrequently assessed. Further, molecular methods, commonly applied in wildlife research, have not been used to examine diet composition of target species in response to diversionary feeding. Here, we used molecular methods to evaluate the consumption of ungulates by American black bears (Ursus americanus) as influenced by the composition of supplemental foods during a diversionary feeding program originally implemented to reduce predation on woodland caribou (Rangifer tarandus caribou) calves in Newfoundland, Canada. We used a previously developed mitochondrial DNA prey identification test to determine the proportion of American black bear scats containing caribou and moose (Alces alces) DNA in 2010 and 2011 at a study site where supplemental foods were provided. The composition of supplemental foods varied between years: bakery waste was provided in 2010, whereas both bakery waste and beaver (Castor canadensis) carcasses were provided in 2011. We also evaluated differences in the consumption of caribou and moose between male and female American black bears at the experimental site and an additional site where diversionary feeding was not implemented. We anticipated that male bears' larger body size and greater nutritional requirements would lead them to more frequently consume caribou and moose. In contrast to our predictions, the proportion of scats found to contain caribou and moose did not differ between years at the site where diversionary feeding was implemented, and there were no differences in the consumption of caribou and moose by male and female bears. This research demonstrates the utility of noninvasive molecular methods to reveal a mechanistic understanding of the effectiveness, or in this case ineffectiveness, of diversionary feeding.