It is nondestructive and rapid method to predict TIN (Total Nitrogen) content of soil in real time using NIR (Near Infrared) spectroscopy.In order to extract the characteristic spectral signature for assessing soil TN content from the soil spectra effectively, analysis of influence factors were investigated and a NIR model predicting TN was developed based on wavelet packet analysis.Firstly, 100 soil samples were collected from the experiment field, and then their NIR spectra were obtained using the high-precision NIR detecting instrument in the laboratory.Secondly, using the orthogonal wavelets of db4 (Daubechies), bior4.4 (Biorthogonal) and sym4 (Symlet) as the generating function respectively, the original signal of each soil sample was decomposed and reconstructed accordingly.Finally, a PLS model for predicting soil TN was established based on the extracted characteristic spectra which were composed of each samples extracted signal using db4, and its determination R2 reached 0.893, and its RMSEC (Root Mean Square Error of Calibration) was 0.0247.The validation R2 was 0.885, and its RMSEP (Root Mean Square Error of Prediction) was 0.0354.And the RPD (Residual Predictive Deviation) of the model was 2.71.It illustrated that the influence factors on TN characteristic spectral signature could be eliminated or substantially decreased from the original spectrum directly or indirectly by using wavelet packet analysis, and the obstacles in establishing linear model for forecasting soil TN were removed, and it is feasible to predict TN content in soil in real-time.