Abstract: In this paper,the correlation between soil spectrum and its Hg content is analyzed on the basis of laboratory measurement and analysis of soil spectrum.Soil samples are collected from the ZhunDong Coalfield,and hyperspectral models for estimating soil Hg content established using Stepwise multiple linear regression.then root mean squared error(RMSE) was introduced to test the predictability and precision of the models,and coefficient of the determination(R²) was used to evaluate stability of the models.Results demonstrate:FDR is optimal index for predicting Hg content.The Stepwise multiple linear regression(SMLR) is the optimal model to establish the relationship between the soil spectrum and Hg content,which R² and RMSEP is 0.93 and 0.036.Estimating Hg content of the soil by the reflectance spectra and improving the accuracy of estimation through varieties of pre-processing methods,so it provides a reference for the recovery and estimate of the ecological environment in study area.