Biochar is often promoted as an amendment to improve soil quality. However, researchers have recently noted that biochar and similar substances preferentially erode from soil, which may reduce its effectiveness. Identifying the erosion mechanisms may help develop strategies for retaining biochar in soil. To investigate the role of raindrop impact biochar erosion, we applied the Hairsine-Rose erosion model to small-scale experiments that simulated rainfall on a simple biochar-soil mixture. The Hairsine-Rose model simulated the biochar concentrations in runoff well for the early part of the experiments but under-predicted the concentrations for longer times. After the simulated rainfall experiments we found that biochar content increased with depth in the soil column from 1% near the surface to 8% in underlying soil layer; similar distributions have been observed for soil, which drives upwards diffusion. By superimposing the Wallach diffusion model on the Hairsine-Rose model we were able to simulate biochar concentrations at both short and long times. We speculate that the relatively dense sand particles are displacing the biochar and we are investigating this further. Our findings suggest that long-term sequestration of biochar in soil is unlikely in soils or parts of the landscape with limited infiltration capacity.