Abstract: The static headspace(HS) combined with gas chromatography-mass spectrometry(GC-MS) was used to analyze the volatile flavor compounds of Zanthoxylum bungeanum from 45 producing areas in Shanxi, Shandong, Sichuan and Gansu. Using automatic mass spectrometry deconvolution qualitative system(AMDIS) and Kováts Retention index(RI) analysis, a total of 99 volatile compounds were obtained. 79, 69, 68 and 63 volatile compounds were identified in Zanthoxylum bungeanum from Shanxi, Gansu, Shandong and Sichuan province. A variety of mathematical models were established to identify the origin. Among them, the Orthogonal partial least squares(OPLS-DA) model could accurately distinguish Zanthoxylum bungeanum samples from 4 provinces, the model fitting parameter Q² was 0.84, and the 6 key difference substances(VIP>2) were piperonone, pinene, terpinene, 3-carene, ocirene and α-phellandrene. Further application of GC-MS full-spectrum data combined with Random Forest(RF) model for origin identification could successfully distinguish Zanthoxylum bungeanum from Shanxi, Gansu, Shandong and Sichuan provinces, with an accuracy rate of 100%.