GC-MS fingerprints of Radix Polygalae (RP) were measured for deliberately collected samples. A total of 88 volatile components were identified and quantified by subwindow factor analysis, heuristic evolving latent projection, and retention index. Next, an efficient discrimination model based on partial least-squares (PLS) discriminant analysis (DA) was developed to distinguish the superior RP samples from the inferior ones, and the reliability and predictive ability of the model was evaluated by cross-validation and permutation tests. Furthermore, four components (1-octanol, shyobunone, isobornyl acetate, and α-asarone) were screened by coefficient β of PLS-DA. They represented the important chemical features of authentic RP and could be applied to the accurate discrimination and QC of RP in the future. Our results suggest that chromatographic fingerprints coupled with chemometric methods provide an effective and convenient strategy for QC of RP and are helpful for revealing the chemical features of a complex analytical sample.