Abstract Based on earlier field cultivation tests and soil investigations, a one-to-one mixing of the second (Aw) and third (B) horizons was conducted to improve the planosol solum in China, leaving the first (Ap) horizon undistributed by a three-stage subsoil mixing plough. The Ap and Aw horizons have a thickness of about 200 mm. The layer below about 400 mm depth is the B horizon. This paper deals with the mechanism of draught production of the second and third plough bodies of the three-stage subsoil mixing plough. The results showed that in both the model and the field tests with the studied soils, the resistances caused by upheaving, tension and cohesion were the largest contributors to the total draught, and the resistances caused by soil acceleration and adhesion were extremely small. In both the model and the field tests, measured and predicted draughts were in reasonable agreement. In the soil bin tests using a half-scale model plough in Japan where a Japanese pseudogley soil was used and the soil hardness in the soil bin was the same throughout, both predicted and measured combined draughts produced by the second and third plough bodies, where each plough body tilled a 100 mm depth, were smaller than that produced when a 200 mm depth soil was tilled by the third plough body alone, if the second plough body was removed and the three-stage subsoil mixing plough was used as a two-stage subsoil mixing plough. In the full-scale three-stage subsoil mixing plough tests in China, the draughts of the second plough body which tilled the Aw horizon and third plough body which tilled the B horizon, were nearly the same, despite the fact that the size of the third plough body was larger than that of the second plough body, because the cohesive and the tensile strengths of the Aw horizon, the second horizon, were larger than those of the B horizon, the third horizon. Here also, the combined draught produced by the second and third plough bodies, where each plough body tilled a 200 mm depth, was smaller than that produced when a 400 mm depth soil was tilled by the third plough body alone, if the second plough body was removed. If the second plough body was eliminated and the Aw and B horizons of the 400 mm depth were directly tilled by the third plough body alone, the draught requirement was steeply increased and large soil clods were formed by the tillage of the third plough body.