The risk of surface-water contamination by herbicides is greatest following application to cropland when the active ingredients are at the maximum concentration and the soil is the most vulnerable to erosion following cultivation. This study determined the magnitude of surface runoff losses of herbicide and nutrients at, and subsequent to, application. The first of three weekly 10-min, 2.6-cm rainfalls were simulated on triplicated 1-m plots (a set) on which corn had been planted and the herbicide (metolachlor/atrazine, 1.5:1.0) and fertilizer (28% N at 123 kg ha−1) had just been applied. Identical simulations were applied to two other adjacent plot sets (protected from rainfall) 1 and 2 wk following herbicide application. Runoff (natural, simulated) was monitored for soil, nutrient and herbicide losses. Concentrations of total phosphorus in surface runoff water and nitrate N in field-filtered samples were not significantly influenced by the time of the rainfall simulation but exceeded provincial water-quality objectives. Atrazine and metolachlor runoff losses were greatest from simulated rainfall (about 5% loss) immediately following application. Subsequent simulated rainfall usually resulted in < 1% herbicide runoff losses. Herbicide concentrations in all plot runoff samples exceeded provincial drinking-water quality objectives. Since herbicide surface transport is primarily in the solution phase (not via association with soil particles), water-management conservation technologies are the key to retaining these chemicals on cropland. Key words: Herbicide, runoff, rainfall simulation, partitioning, water quality