Abstract: In order to estimate the chlorophyll content of tobacco quickly, nondestructively and accurately, the relationship between visible spectral parameters and chlorophyll content parameters was analyzed by designing a two-factor experiment of variety and nitrogen level, and the best parameters were selected to establish the estimation model. The results of nutrient solution culture showed that the interaction between flue-cured tobacco varieties and nitrogen levels increased the variability of chlorophyll and spectral parameters. However, among the three color indices, NRI, R/(G+B), (R-B)/G, (G-R)/(R+G+B), (R-B)/(R+G+B) and ExR were significantly correlated with chlorophyll indices (p<0.01). The correlation coefficients between R/(G+B) and Chl.a, Chl.b and Chl(a+b) were -0.632, -0.636, -0.666, respectively, with the best correlation performance. Field experiments confirmed that R/(G+B) had the highest accuracy in predicting chlorophyll content, and it was chosen as the best color index for estimating chlorophyll content. The RMS errors between R/(G+B) and Chl.a, Chl.b and Chl(a+b) were 0.6069, 0.1567 and 0.7575, respectively. With this method, chlorophyll content of tobacco can be rapidly determined by smart phone, and the fertilization and cultivation management measures can be timely guided. It has a promising application potential.