The near infrared spectra analysis model of the content of the acid soluble lignin and the model of the content of the Klason lignin were built by the iterative method separately at first. The results show that the prediction effect of the content of the Klason lignin is obviously better than that of the acid soluble lignin. Different from usual methods of building near infrared spectra analysis model, the approximate linear relation between the contents of the acid soluble lignin and the contents of the Klason lignin was used. Combined with the near infrared spectroscopy data of multi-wavelength, twenty sub models of prediction of the content of the acid soluble lignin were built with the help of the Klason lignin content whose prediction effect is better than that of the acid soluble lignin. By calculating the weighted mean value of the prediction values of these sub models, the new prediction value of the content of the acid soluble lignin of each acacia specimen was obtained at last. The prediction error of the new model is obviously less than that of the model built by the iterative method. It is possible that the method of modeling in the paper can be used to some chemical component contents when the predictions of them by usual methods are not very effective, and the effects of the near infrared spectra analysis of them will be improved.