Abstract: The study of snow microphysics holds great significance in understanding the mechanism of winter precipitation and optimizing the cloud microphysical parametrization scheme in the weather model. However, at present, very few studies have analyzed the microphysical characteristics of snowfall in China. To thoroughly understand the microphysical characteristics of snowfall in Beijing, this paper presents a detailed analysis of particle size distributions (PSDs), fall velocity (V), snow density (ρ) and axis ratio (ar), and their relationships with equivalent diameter (D) for six snowfall events, using data from a 2D video disdrometer (2DVD). The results are as follows: (1) The density of snow particles with a median volume diameter (D₀) less than 1.5 mm is lower than values reported in other literature. The coefficient in the V(D) relationship is related to density, increasing with it, and the terminal velocity is slightly greater than that observed in Colorado but less than that in Nanjing. (2) The 5-minute PSDs for each case align well with the exponential and normalized gamma distribution models. The temporal variations of the slope parameter (Λ) and the concentration parameter (N₀ for exponential, N_w for gamma) were generally consistent between the two models, while the snowfall rate (SR), total number concentration (N_t) and D₀ calculated from gamma PSD are better consistency with the direct 2DVD measurements. (3) Introducing density (ρ) into the N_w-D₀ relationship improved the correlation. (4) This study presents the first measurements of snowflake aspect ratio for Beijing, showing a mean ar of 0.76 and a tendency to approach 1 for diameters more than 2.5 mm, in agreement with other studies. (5) When the radar reflectivity (Z_e) is constant, an increase in SR is accompanied by a decrease in D₀ and increases in N_w and ρ. Conversely, when SR is constant, an increase in Z_e is closely related to an increase in D₀ and decreases in ρ and N_w. Compared to the Z_e‐SR relationship derived for Nanjing, the relationship obtained in this study (Z_e = 851 SR^1.33) gives lower SR estimates for the same Z_e. These features demonstrate that quantitative snowfall estimation using radar entails greater uncertainty compared to quantitative rainfall estimation.