In this study, soils from 6vegetation restoration modes in the disturbed areas in Xiangjiaba hydropower project were collected, and 25 soil indicators representing soil physical, chemical, and biological characteristics were used to comprehensively evaluate the soil quality by means of sensitivity analysis and minimum data set method. The comprehensive evaluation models of soil quality based on sensitivity analysis and the minimum data set can reflect the impacts of different vegetation restoration modes on soil quality and function differently. The comprehensive evaluation index of soil quality based on the minimum data set under the weighted algorithm (SQI-MWA) can more sensitively reflect the impacts of vegetation restoration modes on soil quality. Organic matter, total phosphorus and polyphenol oxidase can better reflect the differences in the impacts of vegetation restoration modes on soil quality, and are recommended as the best combination of indicators that can reflect the soil quality of study areas in Xiangjiaba hydropower project. The vegetation concrete eco-restoration slope (VC) had the highest soil quality, followed by the thick layer base material spraying slope (TB), natural forest (NF), external-soil spray seeding slope (SS), frame beam filling soil slope (FB), and abandon slag slope (AS). © 2020, Editorial Board of China Environmental Science. All right reserved.