Abstract: The dynamic evolution of ecosystem asset quality (EA Quality) provides a critical basis for scientifically evaluating the effectiveness of ecological conservation. By focusing on temporal evolution trajectories, phase transition characteristics, and spatial heterogeneity, EA Quality assessment can identify trends in ecosystem asset status and quantify the contribution of conservation policies to ecological restoration. As the predominant ecosystem asset type in Sanjiangyuan National Park, grassland plays a key supporting role in maintaining regional ecological security and the supply of ecosystem services. In this study, a multi-dimensional assessment framework for grassland EA Quality was developed, and an empirical analysis was conducted using Sanjiangyuan National Park as the study area. The assessment framework integrates multiple indicators that reflect ecosystem composition, structure, function, and landscape pattern to characterize the status of grassland EA Quality. By combining change rate analysis, spatial gradient analysis, and a multi-zone comparison approach, the spatiotemporal evolution patterns of grassland EA Quality from 2005 to 2023 were quantitatively identified, and the net effect of national park system development was measured. This analytical design systematically revealed the overall temporal trends in grassland EA Quality, determined whether a phase transition in ecosystem asset status occurred after the initiation of the national park system pilot, delineated the spatial spillover boundary of conservation effectiveness, and estimated the independent contribution of the system development process while controlling for the effects of common environmental background changes. The results showed that the grassland EA Quality index increased from 0.4697 in 2005 to 0.5072 in 2023, representing an overall increase of 7.98%. During the background period from 2005 to 2015, grassland EA Quality showed a slight decreasing trend, indicating that grassland was still under some degradation pressure before the national park system pilot. After 2015, grassland EA Quality shifted from decline to recovery, exhibiting a notable phase transition. Meanwhile, the composition, structure, and function indicators of the grassland ecosystem improved in a coordinated manner, suggesting an overall optimization of grassland EA Quality during the study period. Spatial gradient analysis further revealed a clear spillover boundary of conservation effectiveness at approximately 25 km beyond the park boundary. Within this spillover region, the magnitude of recovery rate change of grassland EA Quality was highest among all comparison zones, indicating that the strict conservation and management measures inside the park had a quantifiable cross-boundary driving effect on ecological restoration in adjacent areas. This finding demonstrates that the governance influence of the national park extends beyond its administrative boundaries, manifesting as an identifiable spatial spillover process. According to the multi-zone comparative analysis, after controlling for common environmental background changes, the national park system development had a significant independent net effect on improving grassland EA Quality. This suggests that the grassland ecological recovery trend was not driven solely by common environmental changes, and that the development of the national park system played a substantive role in promoting ecological restoration. These findings indicate that improving the governance system of protected areas requires breaking through administrative boundary constraints and building cross-regional collaborative management mechanisms based on the spatial spillover characteristics of conservation effectiveness. With Sanjiangyuan National Park as an example, this study validates the applicability of the grassland EA Quality assessment framework, providing quantitative support for the evaluation of protected area conservation effectiveness and collaborative governance.