Abstract After 7 years of conventional- and no-till for soybeans on a silt loam soil, replicate plots were tilled alike to produce similar surface conditions for erosion evaluation. Three consecutive rainfall simulations were carried out on each plot to examine whether no-till still had beneficial carry-over effects in reducing soil erosion compared with plots with a conventional-till history. Soil moisture, prior to the first run, was 0.075 g g −1 greater on the previously untilled plots. Their soil loss was less than 30% of that from the previously tilled plots. To eliminate the effect of antece dent moisture content ( M a) on soil loss and isolate the effect of other differences induced by soil use history, an equation between M a and soil loss was used. This had been independently derived from another silt loam soil. Three-fifths of the difference in soil loss were attributed to the greater M a. Two-fifths (4.5 t ha −1) of the difference were attributed to other benefits of no-till such as greater aggregate stability and more biopores. Greater soil loss from the drier conventional-till treatment was a result of the increased breakdown of the soil surface structure before runoff started. The smoother surface resulted in a smaller water depth during runoff. Average depth on the driest conventional-till plot was only one-third of that on the no-till plot with the greatest M a. The reduced depth increased further detachment by raindrop impact and increased sediment transport. The initial breakdown of structure influenced soil loss during all three consecutive runs. For both treatments, the soil loss of the third run was still influenced by the initial moisture content of the first run.