In order to quickly and accurately grasp the hydrolytic nitrogen content of soil in the whole Kunming area,963 soil samples of different types were collected.Partial least squares(PLS) combined with competitive adaptive reweighted sampling(CARS) method used to screen the spectral wavelength variables was adopted to establish an analysis model for hydrolytic nitrogen.Results showed that the parameters for the model were improved after the optimization of wavelength variables by CARS,and the root mean square error of cross validation(RMSECV) was reduced from 31.63 to 25.55,the correlation coefficient of cross validation(Rcv) was increased from 0.78 to 0.84,and the external verification results of the model were basically consistent with the internal cross verification results.Meanwhile,near infrared spectroscopy combined with CARS method could be used to effectively establish a near infrared(NIR) mathematical model for hydrolytic nitrogen content in different soil types in Kunming area under the modeling of a large number of representative samples,which is suitable for the NIR detection of other components in soil and has an important guiding significance.