In order to discriminate fake and genuine cigarettes correctly and rapidly, cigarettes of brand A and fake cigarettes of brand A were scanned by the NIR spectrometer, and an identifying model was developed by near infrared spectroscopy combined with principal component-Mahalanobis distance pattern recognition method. The pretreated spectra data of cigarette samples were analyzed through principal component analysis (PCA), and the result of the analysis suggested that the accumulation of first 4 principal components was more than 97.46%. One hundred samples from total 120 cigarette samples were selected randomly. Then they were used to build qualitative discriminating model and the correlation coefficient was 0.95. Twenty unknown samples were validated by this model. The recognition rate is 100%. The model is reliable and practicable, and could be used as an assistant means for identifying fake and genuine cigarettes.