This study provides a rapid method for quantification of mineral oil in rapeseed oil using near-infrared spectroscopy. The data were processed by direct orthogonal signal correction (DOSC), successive projections algorithm (SPA), partial least squares, and principal component regression (PCR). Good correlation coefficients (R) of 0.998 and root-mean-squared error (RMSE) of 0.005 were obtained, and the DOSC-SPA-PCR model was identified as the optimal method. A satisfactory accuracy with R and RMSE of prediction by DOSC-SPA-PCR of 0.990 and 0.006, was obtained. The results demonstrate that the proposed methodology is a promising method for the rapid quantitative detection of mineral oil in vegetable oil.