application/pdf, To study and model the interrill erosion process in an unmanaged Japanese cypress (Chamaecyparis obtusa) plantation, soil splash detachment rates were estimated based on the quantification of throughfall raindrop indices. Throughfall drops and soil splash detachment were simultaneously observed in the field, and observed data were compared with estimates produced by previous models. Observations took place over five months in 2005, during six observation periods. Raindrop indices of kinetic energy (KE), momentum (M), and momentum multiplied by the drop diameter (MD) were calculated from drop diameters and velocities. The median volume diameter of 1.99 1 mm for the overall observation period was well bounded by those from other Japanese cypress plantations. Throughfall consisted of large drops, generated as drips, exceeding 3 mm in diameter. The fall height was insufficient for the drops to attain terminal velocity, with 91% of the drops reaching less than 90% terminal velocity. The observed throughfall raindrop indices had strong correlation with throughfall rainfall intensity, even though throughfall raindrops occurred in seven rainfall events with different meteorological conditions. The values of observed KE and M were lower than previous model-derived estimations. Earlier models tended to overestimate throughfall KE and M, partly because the expected velocity was greater than that observed, and partly because they did not consider the effect of the splash water component during throughfall. The splash detachment rate in forests was weakly correlated with the total-amount raindrop indices but strongly correlated with the maximum value of raindrop indices over a short time scale such as 1 h. This result indicates that continuous and concentrated raindrop impacts over a short time duration cause splash detachment in the forest floor. Development of a comprehensive model of the process of forest floor soil surface erosion requires more detailed measurement of actual throughfall drops.