Based on near infrared spectroscopy technique and partial least squares (PLS), calibration model of volatile oil content of 141 powder samples with particle size of 40 meshes was established to classify Zanthoxylum bungeagum Maxim more rapidly. After spectra were preprocessed by direct orthogonal signal correction (DOSC), the determination coefficient (R2) increased from 0.936 to 0.95 and the root mean square error of cross validation (RMSECV) decreased from 0.421 to 0.374. Applying the model established by 105 samples to the test set with 36 samples, R2 increased from 0.923 to 0.969, the root mean square error of prediction (RMSEP) decreased from 0.452 to 0.289, RPD increased from 3.622 to 5.573 and RSD decreased from 11.65% to 7.44% after spectra were preprocessed by DOSC. The research indicated that the model preprocessed by DOSC is better, and the precision of model got higher. The result showed that rapid detection of volatile oil content in Zanthoxylum bungeagum Maxim by NIR and DOSC is feasible and efficient. 