[Objective] The particle size distributions (PSD) of different soils measured by scanning electron microscopy (SEM) and by the pipette method (PM) in the water erosion region of China was compared in order to provide a reference for the determination of typical soil PSDs and their applications in soil erosion models.[Methods] Five typical soils corresponding to the five water erosion subregions of China (i.e., black soil, cinnamon soil, loess, purple soil, and red soil) were collected and analyzed by PM and SEM. Specifically, soil suspensions of different size classes obtained by PM (i.e., < 100, < 53, < 20, < 10, < 5 and < 2 μm) were measured using SEM. The PSD results determined by PM and SEM were compared and the causes for their difference were determined.[Results] For most soil suspensions obtained by PM, particles larger than the corresponding size classes of the suspensions were detected by SEM. As the size classes of soil suspensions decreased, both the count and volume fractions of the corresponding sizes, as measured by SEM, decreased. For the suspensions < 100 μm, the particles of the black, cinnamon, and purple soils fell mainly in the 20-53 μm class, resulting in relatively lower values of SEM-derived mean volume diameter (MVD). In contrast, the primary size classes of the loess and red soil were 20-53 and 53-100 μm, leading to larger MVD. Little difference was observed in the PSDs of loess between PM and SEM for the suspension < 100 μm. Compared with SEM, PM overestimated the clay fractions < 2 μm of the other soil samples, and underestimated the size fractions of 20-53 μm for the black, cinnamon, and purple soils, as well as the fractions of 20-100 μm for the red soil. As a result, the PM-derived mean weight diameters (MWD) were smaller than the SEM-derived MVD for all of the investigated soils. Additionally, the PSD discrepancies between the two analytical methods shifted the textural classes of 84.6% of the total soil samples. Converting the PM-derived PSD with the optimal S-curve models, the proportion of the samples with different textural classes due to PSD analytical method decreased to 61.5%.[Conclusion] Based on the SEM results, PM tended to misestimate the PSDs of all of the five typical soils in the water erosion region. The PSD differences between the two methods varied with soil type and were mainly related to soil mineral composition and clay mineral types.