Feral and free-ranging domestic cats (Felis catus) can have strong negative effects on wildlife such as small mammals and birds, particularly in island ecosystems. We deployed camera traps to study free-ranging cats in U.S. national wildlife refuges and state parks in the Florida Keys, and used spatial capture-recapture and kernel density models to estimate cat abundance/density, movement, and activities. We also used stable isotope analyses to examine cat diets captured on public lands. Top models separated cats based on differences in movement and detection with three and two latent groups on Big Pine Key and Key Largo, respectively. We hypothesize that these latent groups represent feral, semi-feral, and indoor/outdoor house cats based on the estimated movement parameters of each group. Estimated densities and activity varied between the two islands, with relatively high densities (~4 cats/km2) exhibiting crepuscular diel patterns on Big Pine Key and lower densities (~1 cat/km2) exhibiting nocturnal diel patterns on Key Largo. Stable carbon and nitrogen isotope ratios from hair samples of cats provided estimates of the proportion of wildlife and human-subsidized foods in cat diets. Cats on both islands consumed mostly human-provisioned foods (>50% of the diet), but some individuals were effective predators of wildlife (>50% of the diet). We provide evidence that cat groups within a population move different distances, exhibit different activity patterns, and that individuals can consume wildlife at different rates. We discuss the benefits of non-invasive sampling and robust statistical techniques to assess a variety of population vital rates.