Abstract: Abstract: Soil erosion is one of serious environmental problems in China. Rill erosion is the most important process of water erosion and sediment production. The sediment yield of rill erosion accounts for most of the total sediment yield. The formation of rill is the result of the interaction between water hydraulic characteristic and soil resistance. Soil particle composition plays an important influence on soil erosion resistance that hinders rill formation and development. The objectives of this study were to clarify the relationship between the rill morphology and sediment yield for soil-sand mixture with different sand contents, which were selected as experimental materials. The indoor water scouring experiment was conducted on the basis of the experimental settings including the experimental combination of 2 slopes (5°, 15°), 4 discharges (2, 4, 6, 8 L/min) and 5 remodeling slopes (sand content of 0, 30%, 50%, 70%, 100%) by a moving steel flume (3 m long, 1 m wide and 0.35 m high). Before the experiment, stones, roots and other impurities were removed from the soil material, followed by air-drying of the soil and sifting through a 5 mm sieve. And the bulk density was controlled at 1.35 g/cm3. Prior to each experiment, the flume was pre-wetted using the 30 mm/h rainfall intensity to moisture saturation with the flume covered with a gauze. The experiment was started when the moisture content reduced to 30% as measured by a soil moisture meter. Mean rill depth, mean rill width and rill width-depth ratio were used as basic rill morphological parameters. The relationship between rill morphology, hydraulic characteristics and sediment yield was analyzed and the prediction equation of slope erosion of different soil particle compositions was established. The results showed that: 1) Mean rill depth increased and mean rill width decreased with the increase of slope gradient, but the impact on discharges was not obvious. The rill cross-section morphology exhibited an obvious regularity that width-depth ratio decreased with the increase of slope gradient and discharges also gradually presented a narrow-deep form．2) Rill cross-section morphology tended to be wide-shallow form with the increase of slope sand content, and rill cross-section morphology of different slope sections tended to be wide-shallow form from up slope to low slope. 3) Unit stream power and stream power had the closest relationship with the rill morphology parameters (r>0.784, p<0.01). There was a non-significant relationship between mean rill width and hydraulic parameters. 4) Mean rill depth and rill morphology comprehensive quantification parameters had satisfying effect on the forecasting of sediment yield (R2>0.747, p<0.01; Nash coefficient was larger than 0.755). 5) The slope erosion prediction equation with preferable reliability was established by the introduction of soil clay content, which facilitated the higher prediction accuracy (R2>0.879, p<0.01; Nash coefficient was larger than 0.887). The results provide a theoretical basis for the study of slope erosion prediction and soil erosion mechanism.