Abstract: Vegetation growth information is mostly represented by hyperspectral reflective curves. In this paper, lots of canopy spectral curves were generated using AvaSpec-4 optical spectrum instrument in order to describe the different growth stage and health status of tobacco. On the basis of analyzing the spectral characteristic parameters such as the red edge position, the area of the red edge, and the height of the green peak, the fractal theory were used to measure the fractal dimensions of the reflective spectral curves in order to analyze the health status of tobacco. The results show that (1) the fractal dimension presents a trend of decrease after the first increase along with the growing of tobacco, and it is in accordance with the change of the three spectral characteristic parameters, (2) fractal dimension has a positive correlation with the three characteristic parameters, and the correlation coefficient is 0.77, 0.91, 0.88 respectively, (3) the fractal dimension of the unhealthy tobacco is lower than the normal obviously. So, the fractal dimension of hyperspectral curve can serve as a new comprehensive parameter to analyze quantitatively the health status of tobacco and provide a scientific basis for growing situation monitoring of tobacco.