In this paper, the reflectance spectral features of chromium(CrCl3), copper chloride(CuCl2) and zinc chloride (ZnCl2) were measured by using visible and near-infrared reflectance spectroscopy (VNIRS). Thus, the spectra features integrating with the state of electrons arranged of heavy metal elements, and combined with the Crystal Field Theory were used to analyze where and why the wavebands of spectral features occurred. With the soil samples collected from Daye City, CrCl3, CuCl2 and ZnCl2 were added to them with different concentrations, and measured their VNIRS to study the impacts of different concentrations of heavy metals on soil reflectance spectra. Besides, the spectral data were transformed via different spectra pre-processing methods to explore the linear correlations between heavy metal concentrations and the reflectance spectra of the soil samples, and to study on where and why the Pearson’s significantly correlated wavebands (p<0.05) appeared and its potential mechanism. The results showed that the reflectance spectra in the range of visible and shortwave near-infrared of heavy metal compounds are related to the whether the 3d orbits of heavy metal elements filled by electron or not. The reflectance spectra of soil samples have been affected by the heavy metal compounds which were added, and negative correlations were revealed between the reflectance spectra data and the heavy metal concentrations. The maximum of the negative values of Pearson correlation coefficient were -0.788, -0.880, -0.824, respectively. There were some changes with the linear correlation between the heavy metal concentrations and the reflectance spectra, and the information of Pearson’s significantly correlated wavebands (p<0.05) became richer and more obvious after the reflectance spectra being pre-processed. This research indicated that the VNIRS of heavy metals are closely related to their electronic structure; the high concentrations of heavy metals in soils could be detected by the VNIRS technique, which has great potential in predicting soil heavy metal contents with rapid and efficient, non-destructive and cost-effective. Based on the VNIRS characteristics of some heavy metals, this paper integrated with the Crystal Field Theory to provide a theoretical basis and experimental reference for qualitative and quantitative analysis of reflectance spectroscopy of soil heavy metals.