As demonstrated in Part II of this paper, the controls on erosion-induced loss in soil productivity are complex. Through the simulation model introduced in Part I, this part examines the biophysical performances of three conservation measures recommended for the Loess Plateau: bench terraces, grass strips and ridge tillage. Bench terraces are found to be the most effective in erosion control. However, the productivity gains in biomass yield saved show grass strips to be the best, followed by terrace and ridge tillage. Significant differences pertain according to initial soil condition and slope. Translating yield increases into monetary terms through the net present value, terraces have the lowest profitability because of their high construction costs. Grass strips have the best profitability. Internal rate of return gave a similar result, while the length of time before break-even is achieved showed that grass strips and ridge tillage become profitable within eight years. An alternative cropping system, now common on the Loess Plateau, of apple orchards on bench terraces was also tested. This gave an internal rate of return of 41 per cent with a payback within seven years, thus proving the attractiveness of this new cropping strategy. Both erosion control and effect on soil quality are key determinants in these analyses, and this research demonstrates how long-term scenarios in a modelling approach can provide good prediction of uptake and performance of conservation measures.