A comparative analysis was done of the SWI (Soil Water Index) acquired by the ERS (European Remote Sensing Satellites) using scatterometers in 1992-2000 vs the data of farmland soil moisture in the Loess Plateau obtained through field investigations and in addition, effects of rainfall, vegetation, land use and irrigation on inversion of the data were analyzed. Results show that inversion of the remote sensed SWI reflects quite well the spatio-temporal variation of soil water in the Loess Plateau, displaying a general trend similar to that of soil moisture in actual distribution in the region, that is, spatially it is relatively high in the south and the east and low in the north and the west, and temporally high in summer and fall and low in spring. Besides, W_swi, soil volumetric water content converted from SWI, are found to be in an extremely significant positive relationship with the measured soil moisture contents in the 0~10 cm soil layer of farmland, which implies that the soil moisture data acquired through inversion of ERS data are very approximate to the actual conditions of near surface soil moisture, and hence the technique can be used to estimate surface soil moisture in the study area. Moreover, high correlation coefficients are found between W_swi and with measured values of soil moisture in the surface soil layer in intensive agricultural plain areas, whereas they are not so high in areas interlaced with farming, forestry and animal husbandry. It is also found in the study that the measured values of soil moisture tend to be higher than those from inversion of the ERS data in relatively dry seasons, mainly due to irrigation that increases soil moisture content in the field. This finding indicates that when using ERS data for estimation of soil moisture contents, it is also essential to take into account the factor of human activities in addition to natural environmental factors, such as landform, soil, vegetation, climate, etc.