Atractylodes macrocephala Koidz. (AM) is a famous traditional Chinese medicine, the quality of AM is directly related to the producing area, so it is urgent to find a method to quickly identify the quality of AM and trace its origin. AM contains various endogenous fluorescent substances, which can provide a basis for identifying the origin of AM by three-way fluorescence technology. In this paper, the three-way fluorescence spectrum of AM was characterized by alternating trilinear decomposition (ATLD), and then three-way fluorescence was combined with partial least squares discriminant analysis (PLS-DA) and k-Nearest Neighbors (kNN) to trace the origin of AM samples. The results show that the classification models established by PLS-DA and kNN could effectively distinguish samples from the three major AM production areas (Anhui, Hunan, and Zhejiang), and the classificationaccurate rate (CCR) of Zhejiang AM, a Dao-di AM herbs, is as high as 80% and 90 in the prediction set, respectively. The proposed method provides a novel and rapid solution for the origin traceability of AM, which is conducive to maintaining the order of AM market and its industrial development.