Runoff simulation tests to investigate the flow mechanics of nonsuomerged overland flow in a natural grass intervening land system were condueted and a modified kinematic wave overland runoff model developed by Choi et al. (1993) was verified. Nonhomogeneity and heterogeneity of the soil, slope, local topography, infiltration, grass density, and the density and activity of the soil microhes and wild animals were the major factors affecting the flow. Streamlines were disturbed by grass stems and small concentrated flows due to the disturbed streamlines and local topography were observed a lot. Relatively larger concentrated flows were observed where bundles of grass were dominant than where individual grasses were growing. Predicted hydrographs were agreed verv well with measured hydrographs. Since the modified model considers grass density in computing flow depth and hydraulic radius, it can be better than existing kinematic wave model if it were used to route nonpoint source pollutant attenuation processes in many grass intervening land systems.