As habitats change, highly specialised species may die or be forced to relocate. However, some obligate coral-dwelling damselfishes appear to survive the localized extinction of their primary habitat, branching coral, caused by coral bleaching. To address this apparent paradox, we documented the spatial behaviour of obligate coral-dwellers in relation to habitat quality. Focussing on two obligate coral-dwelling damselfishes (Pomacentrus moluccensis and Chromis viridis), we used KUDs (Kernel Utilisation Distributions) to quantify fishes’ short-term space use (daily 5 min observations across 6 days) and related it to live coral cover and structural complexity derived from 3D photogrammetry. Specifically, we calculated movement extent (95% KUD), core areas (50% KUD) and the temporal consistency of occupied areas across consecutive days. Structural complexity had no effect on space use. The effect of live coral cover was significant but weak and dependent on fish body-size: core areas increased with decreasing live coral cover for large fishes; smaller fishes showed little response. In contrast to weak habitat effects, there were strong differences across sites. At one site, average core areas increased three-fold to 1.1 m2 for P. moluccensis and over 60-fold for C. viridis, from 1.14 m2 to an average core area of 92.34 m2 and a maximum recorded extent of 1471.4 m2. These findings may help explain these fishes’ apparent, unexpected resilience to habitat loss. Obligate coral-dwelling fishes may prefer branching live coral, but their ‘obligate’ dependence may be more flexible and context dependent. As ecosystems reconfigure, plasticity in fine-scale spatial behaviour may be critical for the persistence of fish populations.