The Jinding giant Zn-Pb polymetallic deposit, located in Lanping County, western Yunnan Province is one of the most important mining areas of lead and zinc resource in China. The continuous active mining has induced increasing heavy soil pollution in and around this area since the 1980s, and is an ideal region to analyze the mining environmental problems. Based on the field work and sample collection, the authors measured the content of four heavy metals of zinc, lead, arsenic, cadmium in 77 soil samples and analyzed the sensitive bands of these heavy metals and their occurrences within soil clay minerals, iron-manganese oxides and carbonates. On such a basis, the Multiple Linear Regression Models (MVLR) was constructed between these four kinds of heavy metals and their measured spectra of the soil. Some results have been obtained: ① the values of the four kinds of heavy metals in this area all exceed the standard values, and show significant correlations; ② the values of these heavy metals have a high correlation with the soil spectral reflectance, the first-order differential equation of the reflectance and the continuum removed variables; ③ the accuracies (R²) of the fitting models are higher than 0.816, which proves that the predicting heavy metal content based on the measured spectra is feasible and reliable. So, the results can provide a technical support for retrieving the heavy metal content of soil by hyperspectral remote sensing technology and for mapping soil heavy metal pollution on large scale by satellite or aerial remote sensing technology.