The objective of the present study was to investigate the feasibility of predicting chlorogenic acid, rutin, scopoletin and total polyphenol in tobacco by Fourier transform near-infrared (FT-NIR) spectroscopy. The partial least squares(PLS) regression method, second derivative and Norris derivative filter were applied in the NIR spectroscopy prediction of chlorogenic acid, rutin, scopoletin and total polyphenol in the range of 7 500 to 4 000 cm-1. For chlorogenic acid, rutin, scopoletin and total polyphenol, the determination coefficients were 0.976 6, 0.941 9, 0.957 1 and 0.966 6, respectively. The SEP/SEC values for them were <1.2, and the SD/SEP values for them were >2. The root mean square error of cross validation (RMSECV) of the four calibration models were 1.938 9, 1.046 2, 0.047 9 and 2.745 2, respectively. NIR spectroscopy was compared with the conventional methods. The results show that the two methods showed no significant difference at the significant level of 0.05. NIR spectroscopy technology can accurately analyze chlorogenic acid, rutin, scopoletin and total polyphenol in tobacco.