Hengshan county of Shaanxi was taken as research area, three kinds of soil organic matter content inversion model based on hyperspectral data were compared. The soil samples were collected in the field. The ASD Field Spec FR was used to measure the soil samples’ spectrum. The content of soil organic matter was measured via potassium dichromate oxidation volumetric method in laboratory. Then the first derivative of spectral data was obtained by applying the reciprocal difference to original spectral reflectance, and the multiple linear stepwise regression (MLSR) analysis model of the first derivative of spectral data was constructed. The correlations between the original spectral reflectance, the first derivative of spectrum and soil organic matter content were analyzed. The first derivative spectra of the characteristic bands which had high correlation coefficient with soil organic matter content were obtained. Based on the first derivative spectra, the MLSR model was established. Meanwhile, the principal component analysis (PCA) was performed for the first derivative spectra of the characteristic bands with high correlation coefficient. The PCA-BP model and PCA-MLSR model were established by the results of PCA. The soil organic matter content was inversed by three methods, and the inversion accuracy was validated and compared with each other. The results showed that the coefficient of determination (R2) and root mean square error (RMSE) between the measured value and inversion value were 0.8930 and 0.1185% with PCA-BP model, respectively, and the R2 and RMSE between the measured value and inversion value were 0.7407 and 0.1613% with PCA-MLSR model, respectively. However, in these MLSR models which based on the first derivative spectra, R2 and RMSE between the measured value and inversion value were 0.6899 and 0.1710% with the optimal inversion model, respectively. Based on the results, the inversion accuracy of soil organic matter content in PCA-BP model was higher than that of MLSR model. In MLSR model, the inversion accuracy of soil organic matter content by using all principal component was better than that only using the partial principal component, of which the cumulative variance contribution was greater than 90%. The content of soil organic matter can be well inversed.