Abstract: Traditionally, the identification of Chinese medicines is performed by the morphological method. As chemical constituents play the role of therapeutic action, it is more rational to identify Chinese traditional medicines by analyzing their chemical constituents. Furthermore, the therapeutic action of a Chinese traditional medicine is usually the result of coordination of its chemical constituents. In this paper, the PRIMA method was adopted to classify 29 samples of the Chinese traditional medicine rhubarb, with 16 samples as the training set and 13 samples as the test set. The recognition ability was 100% and the prediction ability was 92% on HPLC data, while both of them were 100% on UV data. As far as this research is concerned, the PRIMA method is superior to the Bayes classification rule, the SIMCA method and nonlinear mapping in its simplicity, rapidity and correctness. Pharmacologic experiments were carried out to confirm the results.