Prey selection by lions (Panthera leo) reintroduced into small game reserves (typically <1000 km²) in South Africa was compared with results from larger (more naturally functioning) protected areas. Prey selection was used to calculate mean preference indices, and where sample size allowed, tested for significance of preference, as well as for comparisons with current estimates in the literature. The model by Hayward & Kerley (2005) was found to be 85% accurate, suggesting it is well suited as a predictive tool for smaller game reserves. Disparities in prey selection found among study areas, as well as deviations from earlier findings, were highlighted. Potential explanations are offered within the context of optimal foraging theory. Two key issues affecting foraging patterns of lions and which is related to the way reserves are managed were addressed. Firstly, ungulate assemblages not resembling those of natural systems affects the way lions select prey and secondly, temporal declines in the prey base results in changes in lion prey selection. Thus it is not only relative abundance of a prey species that is an important predictor for prey selection, but also overall prey abundance.