The invention relates to a method and an apparatus for predicting a soil light hydrocarbon index. The method comprises: collecting a soil sample from a target area, wherein the same soil sample is divided into two parts, a first soil sample is used for measurement of soil light hydrocarbon content, and a second soil sample is used for spectral measurement; carrying out measurement of the soil light hydrocarbon content on the first soil sample so as to obtain a soil light hydrocarbon contrast value; carrying out spectral measurement on the second soil sample so as to obtain spectral indexes, wherein the spectral indexes include reflectivity, a reflectivity first-order derivative, a reflectivity second-order derivative and a reflectance absorption depth; determining related coefficients by utilizing the soil light hydrocarbon contrast value and the spectral indexes; determining spectral indexes required for modeling according to the related coefficients; establishing a regression model by using the soil light hydrocarbon contrast index as a dependent variable and using the spectral indexes required for modeling as independent variables; and predicting the soil light hydrocarbon index of the target area by utilizing the regression model.