Native mammals across northern Australia have suffered severe decline, with feral cats (Felis catus), introduced herbivores and changed fire regimes being implicated as drivers. However, uncertainty surrounding the relative contribution of each of these threats, and the interactions between them, is limiting the development of effective management strategies. The absence of introduced herbivores and cane toads (Rhinella marina) on Groote Eylandt, Northern Territory, provides an opportunity to evaluate some hypothesised threats in isolation of others. We used camera traps to investigate the correlates of native mammal distribution and abundance at 112 lowland savanna sites across Groote Eylandt. Two large grids of camera traps were also deployed to obtain estimates of feral cat density. We hypothesised that native mammal populations would be negatively associated with feral cat occupancy as well as frequent, large fires. Native mammal site-occupancy on Groote Eylandt was generally higher compared to mainland Northern Territory. Feral cats were infrequently detected, precluding both an estimate of feral cat density and an evaluation of the relationship between feral cats and native mammals. We found no evidence that native mammal site-occupancy or relative abundance is negatively associated with frequent, large fires. The relatively healthy state of native mammal populations on Groote Eylandt is likely due to the low density of feral cats, the benign fire regime and the absence of large introduced herbivores and cane toads. However, due to a lack of historical data, the current state of mammals should not be taken as evidence that these populations are safe from decline. This study highlights that the apparent resilience of mammal populations is a result of complex interactions between factors that vary substantially across the landscape. Caution is therefore required when making broad inferences about the drivers of mammal decline from studies that are spatially and temporally limited.