The spatial and temporal variability of soil water and salt in arid irrigation area is strong. A large number of observation data is often required to monitor, evaluate and predict regional water and salt dynamics. Thus, the selection of temporal stability location is of great significance to the long-term water and salt observations. However, the applicability of the temporal stability analysis methods to soil salinity in irrigated areas, and the similarities and differences of the temporal stability in soil water and salt, as well as the prediction effect of regional salt average with temporal stability analysis are still unclear. Therefore, based on a large number of water and salt observation data in Longsheng study area of Hetao Irrigation District in the Inner Mongolia, the temporal stability feature of regional soil salinity, and their similarities and differences compared with the temporal stability of soil water were analyzed. Then, an improved grouping temporal stability analysis method was proposed, and the prediction effect of this method on regional soil water and salt were verified. The results showed that the lower soil moisture content was with the greater variability, while the higher mean soil salinity was with the greater variability. This was because the heterogeneity of groundwater status and vegetation distribution in spatial location intensified the variation of soil moisture content when soil water decreased gradually. However, for soil salinity, soil salt leaching was mainly affected by soil texture and its own salt content when salt was washed out by sufficient irrigation or rainfall in the whole region. While the process of soil salt accumulation was affected by groundwater conditions, crop types, and so on. The spatial variability of influencing factors in the salt accumulation intensified the variation of soil salinity. Accordingly, the relationship between mean relative deviation (MRD) and standard deviation of the mean relative deviation (SDRD) of soil salt was completely different from that of soil water. The smaller MRD of soil salt was with the smaller SDRD, which was different from the soil water that the smaller SDRD was at the location where MRD was close to 0. The satisfactory results were obtained for estimating regional soil moisture (R² over 0.96) using the minimum SDRD, the minimum index of temporal stability (ITS) and multiple relative time stability points with small SDRD, while there was a large deviation (R² ranging from 0.49 to 0.78) for estimating regional soil salinity using same methods. This is attributed to the differences in relationship of MRD and SDRD between soil water and soil salinity. The soil water and salt data were divided into several groups, and the temporal stability analysis was carried out in each group. The improved grouping temporal stability analysis did not improve the prediction effect of regional soil water, but greatly improved the prediction effect of soil salt with the R² exceeding 0.90. For the long-term soil salinization monitoring locations formed by the improved grouping temporal stability analysis, the change process of soil salt at different salinization levels was considered, which is conducive to understanding the soil salinity evolution process of different salinized soils in time and space.