Abstract: With the rapid development of sponge cities, grass swales have been widely used. Recently, the Chezy formula is generally used to calculate the flow rate along the grass swales. There are differences between the application conditions of the Chezy formula and grass swales, which often cause remarkable calculation errors. One of the reasons for the errors is the Manning roughness coefficient variation. Assuming that the Chezy formula was appropriate, the variation characteristics of the Manning roughness coefficient of grass swales under different constant and variable inflow conditions were systematically researched through full-scale experiments, and the effect of vegetation height on the Manning roughness coefficient was further analyzed. The results showed that the flow rate and Manning roughness coefficient decreased with the increase of flow distance along the grass swale. Under different inflow conditions, the Manning roughness coefficient was closely related to the inflow rate, and the higher the flow rate, the greater the Manning roughness coefficient, with the maximum value being 0.22. In addition, the maximum Manning roughness coefficient of grass swale with different vegetation heights was 0.19-0.22 under different variable flow inflow conditions, and the vegetation height had little influence on the Manning roughness coefficient under the experimental conditions. Therefore, the Manning roughness coefficient of grass swale was markedly affected by runoff volume. The research results can provide reference for selecting the Manning roughness coefficient in the calculation of grass swale drainage capacity.