Context . Source populations of many large carnivores such as tigers (Panthera tigris) are confined within small wildlife refuges in human-dominated landscapes. Appropriate management of these populations may warrant understanding fine-scale use of habitat.

Aims . The aim of the present study is to understand the fine spatial-scale habitat associations of tigers in Chitwan National Park, Nepal.

Methods . We conducted camera-trap surveys across the park and applied an occupancy modelling approach to assess the probability of tiger detection and occurrence as a function of fine-scale habitat covariates.

Results . Tiger detection probability as a function of fine-scale habitat covariates was ≤0.20 compared with that of a constant detection model. Detectability patterns were best explained by models incorporating the effect of prey, slope and landcover type. Similarly, the best occupancy model incorporating the detection probability included prey, landcover type, water and slope. Tiger occurrence patterns were positively associated with prey availability and certain landcover types such as grasslands. Contrary to expectation, occurrence probability decreased further from human settlements. However, as expected, the occurrence of tigers was higher in proximity to water sources.

Conclusions . Both tiger detection and occurrence are influenced by fine-scale habitat factors, including prey availability. In small protected areas, individuals may persist at high population densities by intensively focusing their activity on small portions of their home ranges.

Implications . Our study provided insight into the fine spatial-scale occurrence probability of tigers, and thereby aids in developing appropriate habitat management strategies at the protected-area level. Our approach is broadly applicable to the robust assessment of fine-scale wildlife–habitat associations of many wide-ranging species that are ecologically ‘confined’ in smaller protected areas.