[Objective] To investigate the spatiotemporal evolution characteristics of forest ecosystem health in the mountainous areas of Beijing during 2005—2020，with particular emphasis on analyzing regional variations and the explanatory power of the influencing factors. [Methods] A forest ecosystem health evaluation system was established based on four criteria layers，including geographic environment，vegetation structure，ecological pressure，and vegetation function. The entropy-weight TOPSIS method was applied to calculate indicator weights and evaluate the health of forest ecosystems in the mountainous areas of Beijing. Temporal variations in ecosystem health characteristics were analyzed for the years 2005，2010，2015，and 2020. On this basis， spatial autocorrelation analysis and K-means clustering analysis were employed to investigate the spatial clustering patterns and regional variations in ecosystem health. Furthermore，the single-factor and interactive detection modules of the geographical detector model were utilized to quantitatively assess the explanatory power of various indicator factors influencing the spatial distribution of forest ecosystem health. [Results] 1）From 2005 to 2020，the health condition of forest ecosystems in the mountainous areas of Beijing exhibited a consistent upward trend. At this stage， these forest ecosystems were predominantly moderately healthy（41%） or relatively healthy（48%），forming a spatial differentiation pattern characterized by higher health levels in farther urban areas and lower health levels near the urban zones. 2）The forest ecosystem health in the mountainous areas of Beijing showed continuous improvement，with particularly notable progress between 2015 and 2020，as the proportion of areas classified as ″excellent health ″increased from 2.43% to 18.65%. The Moran's index in the study area exhibited a decline-then-rising trend from 2005 to 2020，indicating significant global and local spatial autocorrelation with clustering patterns. Among the significant spatial types， positive correlations dominated，with HH（41.8%）and LL（30.8%）clusters collectively accounting for 79.6% of all significant spatial types. 3） The spatial heterogeneity of forest ecosystem health in the mountainous areas of Beijing were analyzed at the township level. The results demonstrated that regions such as Huairou，Miyun，and Yanqing exhibited relatively higher ecological health levels，while forest ecosystems in Fengtai， Haidian， and Shijingshan districts showed comparatively poorer health conditions. 4） Soil conservation，Leaf Area Index（LAI），Normalized Difference Vegetation Index（NDVI），tree cover，and population density were identified as the dominant factors influencing forest ecosystem health in the mountainous areas of Beijing. The interactive effects among the factors showed nonlinear enhancement and two-factor enhancement patterns. The primary driving factors of forest ecosystem health in the mountainous areas of Beijing showed distinct variations：GDP and soil conservation were the dominant factors in 2005， followed by LAI and soil conservation during 2010—2015，while air purification and soil conservation were dominant in 2020. [Conclusion] Based on the analysis，the health condition of forest ecosystem in the mountainous areas of Beijing show consistent improvement during the 2005—2020 period，with the role of ecological services becoming increasingly prominent in shaping overall forest ecosystem health. These findings suggest that future forest management and conservation strategies should prioritize the enhancement and maintenance of forest ecosystem service functions.