Vegetation restoration is a key ecological strategy in arid and semi-arid regions, yet its impact on soil moisture sustainability remains controversial. This study takes the Beiluo River Basin on the Loess Plateau as a typical case to evaluate the temporal and spatial dynamics of soil moisture sustainability under ecological restoration. A novel Vegetation-Soil Moisture Suitability Index (VSMI) was developed based on the balance between vegetation water consumption and precipitation replenishment. Multi-source remote sensing and meteorological data from 1990 to 2019 were integrated to conduct comprehensive assessments at temporal, seasonal, and spatial scales. Results show that over the past three decades, the overall VSMI exhibited a significant downward trend (average annual rate:-0.0106 a(-1), P < 0.05), with the proportion of unsustainable areas (VSMI < 1) increasing from 11.74 % to 26.57 %. Forestland showed the most severe moisture deficit, with its VSMI decreasing from 1.218 to 1.117 and the unsustainable area ratio reaching 31.79 %. Clear seasonal variations in VSMI were observed, with summer values (1.250) significantly lower than those in winter and spring (1.343), primarily due to persistent high temperatures, increased evapotranspiration, and reduced precipitation during the growing season. Distinct responses were found among vegetation types: farmland VSMI increased significantly (0.0111 a(-1)) due to improved irrigation and agricultural practices in the northern and central areas; shrubland in the northern loess tablelands maintained moderate sustainability (VSMI = 1.213) but began to degrade after 2000; grassland significantly declined due to water competition from adjacent forestland. Soil texture and topography were key constraints on spatial VSMI patterns, with southern hilly areas exhibiting poor suitability due to sandy soils and shallow soil layers. The proposed VSMI index effectively identifies the water suitability and sustainability of different vegetation types, reveals the hydrological risks and spatial heterogeneity of ecological restoration efforts, and provides a new quantitative tool and decision-making basis for optimizing vegetation patterns and enhancing ecosystem resilience in arid regions.