Forest fires are one of the most significant disturbances in terrestrial ecosystems and have a substantial impact on the biodiversity and carbon cycle of forest ecosystems and pose threats to human lives and property. Due to differences in natural and socio-economic conditions, the main driving factors of forest fires vary across different forest regions. Identifying the spatial pattern characteristics and driving factors of forest fires is crucial for their prevention and management. Focuses on three forest regions in China: Northeast, South, and Southwest forest regions, using the MODIS fire product data from 2000 to 2022 as the primary data source of fire frequency and employing the Geodetector model, this study analyzes the spatial pattern characteristics of forest fire by and exploring and comparing the driving factors of fire frequency and distribution in each of these forest regions. The analysis considers terrain (slope, elevation, terrain position index), vegetation (vegetation type, leaf area index), and socio-economic factors population density, settlement density, proximity to roads, farmland density, road density). The results indicate: (1) Fire points show a clustered distribution pattern from the Kernel Density results, with significant clustering observed in northeastern Fujian and southwestern Yunnan provinces. The driving factors of forest fires vary considerably among different forest regions. In the Northeast forest region, the distribution pattern of forest fires is primarily driven by vegetation characteristics and socio-economic factors, with the leaf area index and proximity to roads having the most substantial impact. In the South and Southwest forest regions, socio-economic factors are the main drivers of forest fires. In the South forest region, proximity to roads is the most significant factor, while in the Southwest, population density and settlement density are the most critical factors. The explanatory power of factors for forest fire patterns is highest in the Southwest forest region, followed by the Northeast, and lowest in the South. The interaction detection results reveal that there are interactions between different driving factors in each of the three forest regions, showing both two-factor enhancement and nonlinear enhancement. This study concluded that the Geodetector model is well-suited for studying the driving factors of forest fires. Given the differences in the driving mechanisms of forest fires across regions, targeted forest fire prevention and management policies should be formulated to provide a scientific basis for fire prevention in China's forest regions.