Abstract: In order to study the feasibility of using soil spectral to predict the content of heavy metal Pb，the paper measured soil spectral data and the Pb element's content of sample points． Using methods of resampling，first-order differential，inverse logarithm，wavelet analysis and envelope spectrum，a stepwise regression model was established to achieve the prediction of Pb element's content through soil spectral data． Regression model based on inverse logarithm processing method was of high precision． Thus，it is feasible to predict the content of Pb using soil spectral．