The aim of present study is to develop a new approach to predicting antioxidative activities of flavonoid extracts from traditional Chinese medicine based on near-infrared spectroscopy (NIR). The 1,1-diphenyl-2-picryl-hydrazyl (DPPH) method was employed to assay the antioxidative activities of twenty-eight extracts. Then, the near infrared diffuse reflectance spectra of those samples were acquired in the range of 4 000-10 000 cm-1. The partial least square (PLS) algorithm was used to generate the calibration model that correlates the spectra and the antioxidative activities of those natural products. The optimal wavenumber ranges and the preprocessing method of original spectrum data were selected during the establishment of calibration model according to the root mean square error of cross-validation (RMSECV). The established model has been successfully applied to predict bioactivities of six samples in the validation set. The value of RMSECV of the proposed model was 9.50% with R2 of 0.901 7, and the value of RMSEP was 14.8%. The result indicated that the method can be used for the fast determination of the antioxidative activities of flavonoids as well as other active components of natural products.