Abstract Simulated rainfall was applied to a Tama silt loam soil under four tillage treatments, two crops (corn and soybeans), two crop stages (at planting and one month later), and two row orientations (on the contour and up-and-down the slope). Contouring significantly increased the quantity of rainfall applied before the initiation of runoff for all tillage treatments. However, after a drainage pattern was established, contouring had little effect on runoff. Depth of runoff was usually greatest from moldboard-plowed plots and least from the strip-till plots, and differences between these two treatments were usually statistically significant at the 95% level of probability. Runoff depths from the chisel-plowed plots were usually less than the moldboard-plowed plots but the differences were never statistically significant. Runoff depths from the no-till plots were often significantly less than the moldboard-plowed plots, particularly in the year following soybean production. As the soil approached saturation, steady runoff rates were an average of 7 mm h −1 greater when the previous crop was soybeans for all tillage treatments, except for the no-till. Greater runoff following soybeans could be partially explained by less soil residue following soybeans, since runoff depths and steady runoff rates from the tilled soils were negatively correlated with residue cover. The relationship between residue cover and runoff for the no-till following corn appeared to be different from all other treatments. The least runoff rates were observed from plots which had been recently tilled yet retained high levels of soil residue cover.