Abstract: To investigate the spatiotemporal evolution characteristics and scale differences of ecosystem health from a land use evolution perspective, this study evaluates regional ecosystem health based on land use in the Chengdu-Chongqing urban agglomeration. It reveals its spatiotemporal evolution, multi-scale spatial clustering distribution patterns, and driving mechanisms, while projecting future ecosystem health under various scenarios using the Land Use Simulation PLUS model. Results indicate that from 2005 to 2020, the territorial spatial pattern of the Chengdu-Chongqing urban agglomeration primarily exhibited a “one increase, two decreases” trend: continuous expansion of construction land alongside persistent declines in cultivated land and grassland areas. Ecosystem health demonstrated significant spatial clustering with pronounced scale differentiation, where mean values generally followed the order grid > county > grid cell. Single-factor analysis identified land use type, DEM, annual mean temperature, and slope as primary drivers of ecosystem health. Two-factor interaction analysis revealed that regional ecosystem health is jointly influenced by natural and socio-economic factors, with the interaction between land use type and slope further amplifying the driving effect. Multi-scenario projections reveal that across all development scenarios, grid-scale ecosystem health averages outperform both raster and county-level scales. Moreover, under the ecological conservation scenario, ecosystem health averages at all scales exceed those of the natural development and urbanisation scenarios. This indicates that the ecological conservation approach is most conducive to regional ecosystem health and sustainable development.