Abstract: The study area is Dianchi Lake Basin, the spectral reflectance of 4 typical wetland plants, including Phragmites australis, Zizania latifolia, Typha orientalis and Cyperus involucratus, was acquired by S-R3500 spectrometer, nitrogen content of wetland plant leaves was determined by model A33 of continuous flow analytical instrument, the most sensitive spectral bands and spectral transformation formal parameters in leaves of 4 wetland plants with nitrogen contents were selected, and the nitrogen contents estimation models of wetland plants were established with hyperspectral data. The results revealed that the spectral curves of 4 wetland plants were consistent with those of the healthy green plants, there are differences in the wavelength of 800–1350 nm and 1650–1800 nm, the spectral characteristics of 4 wetland plants can be more highlighted through transformation of spectral includ first-order differential, second-order differential and "three-edge" parameters; the highest absolute correlation coefficient between wavelength and nitrogen content is found in the second-order differential band spectrum of P. australis(1,836nm), the first-order differential band spectrum of Z. latifolia(1264 nm), the second-order differential band spectrum of T. orientalis(881 nm), the second-order differential band spectrum of C. involucratus(1665 nm). The correlation of the "three-edge" parameters that ratio of red-edge and blue-edge area and the normalized value of red-edge and blue-edge area of Z. latifolia, the red-edge amplitude, red-edge area, the ratio of "green peak" and "red valley" reflectance and ratio of red-edge area and blue-edge area of T. orientalis with the nitrogen contents were the largest and significantly correlated(P<0.01); the accuracy of nitrogen content estimation models of wetland plant leaves based on spectral second-order differentiation was higher than that of first-order differentiation, the nitrogen content estimation models based on the "three-edge" of the spectral parameters had the highest accuracy and the least error; the estimation results for nitrogen content based on "three-edge" parameters of 4 wetland plants are good using a multiple stepwise regression estimation model. The estimation models techniques of wetland plants of nutrient contents based on hyperspectral technology can rapidly monitor the spatio-temporal dynamic changes of nutrient contents of large-scale wetland plants in a long time series, providing basic data and technical support for remote management of wetland plants.