Abstract: Abstract: Soil erosion newly derived from engineering spoilbank severely threatens ecological safety of construction sites and surrounding areas. Engineering spoilbank with three-dimensional is quite different from the traditional slope, as a result, to study the hydrodynamic process of a three-dimensional engineering spoilbank is of great necessity. In this study, in order to find the best hydrodynamic parameters that can describe the hydrodynamic process of soil erosion caused by tapered engineering spoilbank, laterite collected from the north area of Jiangxi province was selected as the main test material and variations in hydraulic parameters (flow shear stress, stream power, unit stream power, unit energy of water-carrying section) of overland flow were studied experimentally. This study was processed in inner-door rainfall experimental condition. Three kinds of artificial intensities (1.0 mm/min, 1.5 mm/min, 2.0 mm/min) and three gravel content (10%, 20%, 30%) were designed. Every experiment persisted 33min when there was no erosion gully occurring on the surface of the tapered engineering spoilbank. The result shows that: 1) instantaneous value and average value of independent hydrodynamic parameter were closely related to soil detachment rate except the cross-section specific energy. To be more specific, both the flow shear stress and the instantaneous flow shear stress had a significant power function relation with the corresponding detachment rate and this relationship also occurred between the stream power and the detachment rate, while, the relationship between unit stream power (or the instantaneous unit stream power) and the detachment rate (or the instantaneous detachment rate) was linear. In terms of the cross-section specific energy, its instantaneous value had no obvious relations with the instantaneous detachment rate though its average value had a power function with the average detachment rate; 2) critical unit stream power for surface erosion of the tapered spoilbank was 6.8×10-3 m/s, and the critical instantaneous value of surface erosion parameter tended to be less 3.8×10-3 m/s. Moreover, the erodibility parameter of experimental material was about 0.0053-0.0059 s2/m2, and that was 20-30 times larger than the pure clay. So, soil mixed with gravel on the spoillbank surface was more easily carried away by the overland flow than the pure clay with no stirred; 3) the relevance between the instantaneous cross-section specific energy and the instantaneous detachment rate was extremely low (R2=0.130), so instantaneous cross-section specific energy was not suitable to be a hydrodynamic parameter describing the hydrodynamic process of soil erosion caused by tapered engineering spoilbank; 4) according to values of fitting effect, the parameters tested should be ranked in the order of stream power>instantaneous stream power>unit stream power>flow shear stress> instantaneous unit stream power>instantaneous flow shear stress>cross-section specific energy, obviously stream power had the best correlation (power function) with the soil detachment rate (R2=0.972). So stream power was the best one of all the hydrodynamic parameters tested to describe the hydrodynamic process of soil erosion caused by tapered engineering spoilbank. Stream power tends to be more scientific to study the erosion process of tapered engineering spoilbank on the energy hand. In summary, the hydrodynamic process of a tapered engineering spoilbank is quite different from that of a slope. The result obtained may lay the foundation for the establishment of process-based model of soil erosion caused by engineering spoilbank, and provide a scientific guidance for comprehensive control of soil and water loss in production and construction areas.