In order to quickly and effectively control the quality of tobacco flavors of different batches, the volatile components in the samples from six batches of the same flavor with the inclusion of 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40% (mass fraction) other flavors and diluted with 5%, 10%, 15%, 20%, 25%, 30%, 35%, 40% (mass fraction) ethanol, and two other tobacco flavors were analyzed with simultaneous distillation and extraction and GC/MS. The fingerprint spectrums of the samples were established by calculating the relative retention values and relative peak areas of all chromatographic peaks of those volatile components, and the pattern recognition of flavor samples was carried out with principal component analysis (PCA) and projection analysis of the fingerprint spectrum. The results showed that flavor sample with the inclusion of 5% other flavors or 5% ethanol was distinguishable by PCA projection method. PCA projection method can be used as an effective means for quality control of tobacco flavors.