A method to determine some physical and chemical indices of tobacco flavor and fragrance simultaneously and rapidly was established. Prediction models of four indices including refraction index, pH, relative density, and total volatile constituents were founded by using Fourier transform near-infrared (FT-NIR) spectroscopy. The effect of spectra pretreatment on models was studied. At the same time, reliability and repeatability of the predicted results of these models were evaluated. Results show that these models are applicable to different flavors and fragrances, and can predict simultaneously four indexes including refraction index, pH, relative density, and total volatile constituents. The values of RMSEP are 0.00182, 0.374, 0.00393 and 3.04, respectively. Therefore, the qualities of tobacco flavor and fragrance can be controlled rapidly without any damage by FT-NIR.