Taking Feng-qiu County as a case of soil salinization widely existing in the semiarid region, the spatial variability of soil salinity was investigated by using remote sensing and EM (electromagnetic induction) technologies in the present study. Descriptive statistics was applied to soil salinity data interpreted from EM38 measurements using field sampling method. Spectral indices (soil index and plant index) were derived from 25-resolution Landsat TM image taken in April 2005, and proved to be significantly correlated with soil salinity interpreted by EM38 readings. Regression models were further established between the interpreted soil electrical conductivity and spectral indices (soil index and plant index), and spatial distribution patterns across the study area were finally mapped based on the above regression models. Results indicated that soil salinity at each soil layer is from 0.259 to 0.572 and exhibits the moderate spatial variability owing to compound impact of intrinsic and extrinsic factors. Spatial distribution maps of soil salinity were obtained with the application of plant index, soil index and EM38 measurements. It was shown that soil salinization, mainly located in the north and south of the study area, exhibited obvious trend effect. Salinity at surface soil was the greatest and showed the trend of a decrease at subsoil layer and then an increase at deep layer in the whole soil profile. The accuracy of the predictions was tested using 40 soil sampled points. The root mean square error (RMSE) of calibration for soil salinity in each layer was 0.094, 0.052, 0.071 and 0.067 ds·m-1 respectively, showing that the precision is ideal. The change trends of RMSE were the same as soil salinity in soil profile. The trends indicated that soil salinity had effect on the salinity prediction by spectral indices, and showed better accuracy at low soil salinity.