Discriminant analysis is an important method in multivariate statistic analysis to distinguish what type an individual should belong to. Based on the content of total sugar, reduced sugar, total nitrogen and alkaloid, the discriminant analysis was used in the recognition of different qualitative flue-cured tobacco. At last, the discriminant function was developed. The rate of accuracy, of returned classification using methods of mutual test, reached to 91.43%. The rate of accuracy, of returned classification using new samples, reached 100.00%. The results showed the method of discriminant analysis was useful and the discriminant function was valid.