Satellite polarimetric observation is an important approach for remote sensing of atmospheric aerosols. The validity and adaptability of aerosol type model is one of key factors affecting polarimetric remote sensing of aerosol properties. For an aerosol retrieval algorithm, either neglecting coarse mode aerosols (one of common assumption of aerosol retrieval from satellite polarimetric sensors) or employing incorrect aerosol types can bring errors to the retrieval results. In this paper, the influence of aerosol modes and choice of aerosol types (six typical types: desert dust, biomass burning, background/rural, polluted continental, polluted marine, and dirty pollution) on the retrieval of aerosol optical depth (AOD) were investigated based on near-infrared polarized satellite channel at 865 nm. Based on the vector radiative transfer simulation results, the atmospheric bidirectional polarized reflection distribution functions (BPDF) of the above six aerosol types at 865 nm were analyzed. The results demonstrate that atmospheric BPDF depends greatly on aerosol particle’s size, while influence of coarse mode on atmospheric BPDF is much smaller than that of fine mode. Compared with the atmospheric BPDF associated with only one fine mode, the BPDF associated with both fine and coarse modes shows smaller BPDF. With these simulation results, AOD retrieval errors were analyzed under two circumstances, neglecting coarse mode aerosols and employing an incorrect aerosol type. The results show that: (1) Neglecting coarse mode can lead to underestimation of the retrieved fine mode AOD (AODf). The AODf errors are in the range of -12.3% to -35.7% for different aerosol types. The maximum AODf error associates with desert dust aerosol type while the minimum with polluted continental type. (2) The use of an incorrect aerosol type can bring large AOD retrieval error, which depends on bias of atmospheric BPDF related to the incorrect aerosol type. Among all six aerosol types, the maximum BPDF difference occurs between desert dust and dirty pollution aerosol types, which yields the maximum AOD retrieval errors of -60.6% and 220.3%, given mischoosing one versus another. The minimum AOD retrieval errors can be -3.0% and 7.1% occurring between background/rural and polluted continental types, which have the smallest BPDF difference. The results of this study can help to develop the next generation spaceborne polarimetric sensors and corresponding aerosol retrieval algorithms.