One of the main factors limiting the acceptance of large carnivores is livestock depredation. Reducing damages on livestock requires understanding how depredation varies in space and time. The conservation of the brown bear (Ursus arctos) population in the Pyrenees offers a relevant study case to illustrate this issue, with a minimum population size of 41 individuals recorded in 2016 and an average of 103.3 ± 18.9 attacks per year on domestic animals between 2010 and 2016 during the summer pasture period. We analysed the spatial aggregation of depredation events by using the local Getis-Ord analysis of spatial dependence at the management scale (pastoral units) and at a finest scale (250 × 250 m grain). Our results uncover the absence of coldspots of brown bear depredation in the French Pyrenees and the presence of significant hotspots. Depredation hotspots are consistent in time, meaning that a hotspot in one year is likely to exist in the following year(s). The fine scale analysis allowed identifying both inter-and intra-pasture hotspots and we propose a simple method to rescale these fine scale results. We linked this spatial pattern of hotspots (at 250 m resolution) to environmental factors. Hotspot presence is characterized by being close to forest and buildings with a high proportion of grassland and on steep slopes. Moreover, a nonlinear relationship with brown bear activity describes the presence of hotspots. The assessment of depredation hotspots and their link with environmental factors offers some practical guidance about where to focus efforts in order to decrease this human-large carnivore conflict.