A pot cultivation experiment was carried out to investigate the relationship between uranium contents in leaves of Phytolacca acinosa Roxb. and original spectral datum and first derivative spectral datum with derivative technique, and to seek out the sensitive wavelengths and spectral characteristics of Phytolacca acinosa Roxb. under uranium pollution. Then by choosing sensitive bands and the best correlated spectrum characteristic parameters, uranium estimation models were constructed. The results showed that when the U contents in leaves were 5.94～71.74 mg·kg-1, they correlated closely with the first derivative reflectance in the range of 749～766 nm. The chosen 14 spectral characteristic parameters were used to calculate the correlation coefficients with uranium contents in leaves, and correlations of the blue edge area, the red edge position, the ratio of red edge area to blue edge area and the normalized values of red edge area and blue edge area were significantat the 0.05 level. The selected wavelengths of 757, 758, 760, 761 nm and the above-mentioned 4 best spectral characteristic parameters were used to establish the uranium estimation models, and precision tests proved that the uranium estimation models established bythe ratio of red edge area to blue edge area, first derivative reflectanceat 757 and 760 nm achieved better test results, among them, the best model was the cubic function model using first derivative reflectance at 757 nm as a variable and the prediction accuracy of it was up to 89.8%.