Globally, wildlife managers often control predator populations to protect biodiversity, livestock or other valued resources. Most assume that the predation impact of each individual predator is the same and that removing any individual predator produces a benefit to the target species. However, research suggests predation efficacy can vary within a predator species according to phenotypic characteristics. Understanding these individual differences may be critical for managing predation impacts on particular categories of prey including small populations where predation effects are amplified. We used dietary data from 1748 feral cats euthanized over 30 years during a control programme in arid Australia to determine whether any predator attributes could predict their effect on different prey weight classes. Feral cats in our study ate a wide range of prey including reptiles, mammals, birds and invertebrates. Demography (body mass) was a highly significant predictor of diet. Cats weighing 3 kg fed predominantly on prey <50 g increasing to >500 g when cats attained a body mass of 6 kg. Of more significance was that diet varied between demographic groups but also within a single demographic group (adult males) based on body mass, with results having significant implications for threatened prey. Modelling indicated that for a fixed predator population size, the predation rate on prey in a >500 g weight class increased by 28% when the ratio of large (>4.2 kg) to small adult males varied according to its natural range (12%–80% over the 30-year study). Results suggest that variations in predator demography can significantly impact predation rates on prey species and should be included in predator–prey models for small prey populations. On average, large male cats comprised 23% of the population and our findings suggest that targeting this ‘lethal demographic’ and manipulating predator demography should be prioritized along with lowering predator density to reduce predation impacts on prey weighing more than 500g.