Abstract: The hyperspectral remote sensing technology possesses a huge potential in studying soil components because of its high spectral resolution, strong waveband continuity and rich data. The HSI (Hyperspectral imager) was independently developed by Chinaand carried on the resources and environmental satellite, and it has many characteristics, such as the wide spectrum with “integrated image and spectrum” (0.45-0.95 μm), long return cycle (96 hours) and detectability of fine spectrum (5 μm). However, its data application is still in exploration in China. Taking the HSI hyperspectral remote sensing image as a data source and combining with the test results of salt ions in soil, [WTBX]EC[WTBZ] and pH of field soil samples were analyzed so as to reveal the spectral features of salt ions in HSI data by using the geographic information system (GIS) and partial least square regression (PLSR), a PLSR model of soil salt content was developed based on the hyperspectral data, and a retrieval experiment on the main parameters of soil salt content in the Songnen plain was accomplished. The research result indicated that PLSR could be used to decrease the dimension of spectral data and increase the analysis efficiency under the premise of guaranteeing the maximum information amount. The prediction model developed using PLSR could be used to achieve a good retrieval accuracy of total salt content, [WTBX]EC[WTBZ] value, Na++K+, Cl- and HCO-3, and their determination coefficients reached 0.799, 0.879, 0.772, 0.791 and 0.694 respectively. In the quantitative retrieval of soil salt content, the HSI image was applied as a new data source to provide an important guidance in acquiring the values of soil salt content and prevent soil salinization in the Songnen Plain accurately, quantitatively and quickly.