While vegetation greens, substantial water consumption by ecological restoration (ER) has attracted increasing attention. China has become a major contributor to global greening through great ER efforts; however, the country's water scarcity and highly uneven distribution of water resources may imply spatially varying effects of ER on regional water security, though this remains unknown. Here, by providing comprehensive satellite evidence to depict the ER process and water storage changes, we systematically assessed the risks posed by ER practices to water resources in China. We showed that changes in forest areas and related ecological indicators were highly consistent, with the strongest ER efforts occurring along the route from Southwest to North to Northeast. Comparing the spatial patterns of ER-induced water loss through ET and GRACE satellite-derived water storage depletion, North China, one of the country's largest granaries with the most severe water scarcity, was identified as the hotspot experiencing the most significant ER process and the rapidest decline in terrestrial water storage (TWS: -14.65 mm/yr) and groundwater storage (GWS: -17.06 mm/yr). Quantitative attribution analyses, integrating water balance analysis and statistical models (e.g., Structural Equation Model), revealed that rising ET was the primary factor responsible for TWS decline in the ER regions. We suggested that the gravest threat to water resources from national ER occurs in North China, exacerbating the water shortage originally induced by agriculture. We caution that assessments of water-carrying capacity should be incorporated into the formulation of future ER strategies in water-scarce regions, particularly in North China, a region of great strategic significance. This study can not only provide critical insights for optimizing the afforestation layout to ensure the synergy between ecological and water security and promote sustainable development, but also offer typical cases for other developing countries to avoid similar dilemmas.