In order to research the adaptability of the NIR quantitative calibration models for the metal elements in soil, in this research, near-infrared spectroscopy combined with partial least square regression algorithm was applied to develop the quantitative calibration models of K, As, Hg, Cu, Zn, Pb, Cr, Cd in the air-dry soil samples after the outliers having been eliminated. The content prediction of the elements mentioned above was carried out for the air-dry soil samples and the oven-dry soil samples for the external validation set respectively. The result indicates that the correlation coefficient between the estimated and specified values of the air-dry soil samples is larger than that of the oven-dry soil samples for each element. A significant correlation exists between the estimated and specified values of the air-dry soil samples for each element, while there is no significant correlation exists between that of K, Hg, Cr of the oven-dry soil samples. In this thesis, the adaptability of the NIR quantitative calibration models for the metal elements in soil was researched preliminarily, which, to some extent, can provide reference for the rapid quantitative monitoring method of the metal elements in soil and the monitoring of the home environment of agricultural products.