Abstract: 120 bitumen samples from 6 different manufacturers were collected and analyzed by attenuated total reflectance Fourier transform infrared spectrometry. The spectra data were analyzed by pretreatment method and multidimensional scaling (MDS). The MDS scatter plot revealed that the bitumen samples were distributed in 6 different areas, and samples appeared in the same area were the bitumen from the same manufacturer. Based on the data of MDS obtained from a set of known sample, regression model was established, and could be used to identify an unknown sample in the MDS scatter plot by substituting its data into the model.