Nearinfrared (NIR) and midinfrared (MIR) spectroscopy are powerful tools for quantitative analysis of material composition due to its rapid, nonpolluting, nondestructive characteristics. The two infrared spectra show different characteristics. The advantages and shortcoming of the two spectra were analyzed and compared in the view of the mechanism, spectral analysis models and spectral instrument designs of the two spectra. This provided a good method for the selection of suitable spectroscopy. The hydrocarbon source rocks were analyzed by both midinfrared spectroscopy and nearinfrared spectroscopy. The wavelet multiscale direct orthogonal signal correction (WMDOSC) method was used to preprocess original NIR and MIR spectra. The results show that this method can significantly improve the accuracy and stability of the nearinfrared spectral analysis model, and the results of nearinfrared spectral analysis model are consistent with that of midinfrared one.