Soil salinization in Ugan and Kuch river delta, located in the north part of the world’s second largest desert of Taklimakan, has become one of the main environmental problems for regional sustainable development with the rapid development of the economy and the continual utilization of soil resources. How to get the precise information on soil salinization is very important and useful to monitor soil salinization and its dynamic changes. This paper, by combining the regional background-information and based on the full analyses of spectral features of RS data, revealed the brightness changes of TM images and its separation regulation with PCI software for soils under different salinizated-extents in Ugan and Kucha river delta. The results showed that the total extracting accuracy of salinizated soil was 88.6%, and those of severe- and middle-light salinizated soils were 90.8% and 86.7%, respectively. In conclusion, the method used in this paper could extract accurately the information on salinizated soil.