Abstract: Taking Sentinel-2 A remote sensing images as test materials, the total phosphorus content of bare soil in Fuyu city was studied by using decision tree classification, correlation analysis and stepwise regression analysis, in order to provide data support for the implementation of regional fine agriculture and provide reference for the study of Sentinel-2 A remote sensing data in soil composition detection. The results showed that the decision tree analysis method could effectively separate the bare soil in the study area; the reflectance of Sentinel-2 A remote sensing image had a good correlation with the total phosphorus content of bare soil in the study area after the first-order differential transformation of band quotient and band reflectance reciprocal, and the correlation coefficients reached-0.826 and 0.831 respectively; the inversion model of total phosphorus content of bare soil established by quotient of sentinel-2 A band was Y=-613.5X_（3/2）-545.0X_（12/2） +3 679.1X_（8A/7）+494.9X_（11/8）-3 609.2,the determination coefficient of the model reached 0.780,and the root mean square error was 113.1 mg·kg^-1,indicated that the method of retrieving total phosphorus content of bare soil in Fuyu City by using Sentinel-2 A remote sensing data was feasible.