The Ecological Water Conveyance Project (EWCP) is an important measure to restore the degraded vegetation ecosystem in Qingtu Lake Wetland (QLW), a terminal lake wetland of an arid inland river basin in northwestern China. Although many previous studies have demonstrated the positive effect of EWCP for vegetation restoration, they have ignored the possible soil salinization by EWCP which could lead to re‐degradation of restored vegetation in QLW. Therefore, the purpose of this study is to reveal the potential risk of soil salinization on sustainable vegetation restoration in QLW. Three satellite remote sensing‐based indices (i.e., Normalized Difference Vegetation Index (NDVI), Salinity Index, and Normalized Difference Water Index) are used to indicate the spatio‐temporal evolutions of vegetation growth status, surface soil salinity, and water surface. Additionally, ground data of groundwater, soil, and ecological water volume are collected. The development of soil salinization and its risk on vegetation restoration as well as its controlling factors are subsequently studied by combining remote sensing and ground‐based data. The results indicate that the vegetation restored during the first few years of EWCP in the southwest, west, and north of QLW has degraded again due to the aggravation of soil salinization in these regions since 2019. Soil salinization was accelerated by the low groundwater depth with high mineralization and high evaporation capacity of climate without adequate inundation of ecological water. Under current ecological water management in QLW, soil salinization will intensify and vegetation will further degrade. The artificial management of ecological water in QLW has created a complex relationship among vegetation growth, soil salinization, and groundwater. This indicates the irrationality of using a fixed relationship between groundwater depth and NDVI for the computation of optimal ecological water volume in previous studies. More field monitoring of groundwater‐soil‐surface water‐plant‐atmosphere continuum and process‐based modeling of water‐salt transport and vegetation growth should be conducted in the future.