Abstract: In order to study the feasibility for a rapid method to calculate the contents of free amino acids in tobacco by Fourier transform near-infrared (FT-NIR) spectroscopy, the NIR spectroscopy models for calculating the contents of histidine (His), asparagine (Asn), glutamine (Gln), proline (Pro), phenylalanine (Phe), tryptophan (Trp) and the total content of free amino acids in tobacco were developed. The approach used a partial least squares (PLS) regression method to model and the first derivative and Norris filter to pretreat obtained spectra in the range of 4 000-9 000 cm^-1. The results from the NIR spectroscopy method and conventional standard method were compared, and the significance of deviation of the developed models was verified. The results showed that: 1) For the optimal PLS prediction models of the said amino acids and the total content of the free amino acids, the correlation coefficients (r) were higher than 0.9, the ratios of standard error of prediction (SEP) to standard error of calibration (SEC) were less than 1.2, and the ratios of standard deviation (SD) of verification set samples to SEP were higher than 2. 2) The root mean square error of cross validation (RMSECV) of the seven calibration models for the contents of the selected amino acids and the total content of free amino acids were 0.021 7, 0.186 0, 0.202 0, 0.818 0, 0.032 6, 0.047 8 and 1.592 0, respectively. 3) There was no significant difference between the two methods at the significant level of 0.05. Therefore, the NIR spectroscopy technology can be used to accurately quantitatify the contents of these amino acids and the total content of free amino acids in tobacco.