The maintenance of biodiversity is a global concern in economic, social and environmental terms. Thus, conservation units for the protection of natural environments were created. Despite the importance of these areas, forest fires have caused immeasurable and constant damage to Conservation Units. In view of this, the objective of this study was to determine the risk areas of forest fire occurrence through Fuzzy logic modeling in the Córrego Grande Biological Reserve, located in the Mata Atlântica Brazilian biome. In order to prospect for areas at risk of forest fires, the following variables were inserted in the model: land use, road network, slope and relief orientation. Finally, the model was validated by comparing the location of fire occurrences between the years 2008 and 2018, and a layout of the risk classes in the study area. In doing so, it was found that 65.87% of the area is between the 'moderate' and 'very high' range of fire risk classes, and that 70.22% of the fires which occurred in the studied period also occurred in that class range. The study concluded that the most influential variable on the risk level of fire occurrence is the forest road network. In this way, the proposed methodology can be applied to any other areas and types of land cover.