Subsurface versus surface application can reduce herbicide losses from surface runoff, volatilization, photodegradation, and wind-induced drift. Because distribution in the soil plays an important role in herbicide fate and transport, this study was conducted to analyze the effect of various application methods on herbicide losses with surface runoff. Twelve rainfall simulation plots (three replications of four herbicide application treatments) were established in 1995. Losses of atrazine, metolachlor, and cyanazine with surface runoff were measured for the four different treatments: broadcast spray without incorporation with no-till (NT), broadcast spray with disk incorporation (SD), broadcast spray with Mulch Master incorporation (MR), and subsurface application with incorporation using a modified Mulch Master (MB). For the modified Mulch Master, sprayer nozzles were added to the trailing edges of 61-cm wide Mulch Master sweeps, which were run at a depth of 6 cm. Following herbicide application, rainfall was simulated at 6.35 cm h ^-1 for 1.5 h on the 3.1 m × 10.7 m plots. Runoff volumes and soil losses were greatest for NT (3.36 cm and 632 kg ha ^-1 ), followed by MR and MB, with SD showing the lowest total runoff and sediment losses (0.34 cm and 217 kg ha ^-1 ). Herbicide losses and concentrations were significantly greater (P = 0.10) for NT than for the other three treatments. On the NT plots, runoff began quickly and only occurred on two of the four inter-row areas that had traffic tracks. Herbicide losses decreased in the order of NT, MR, SD, and MB. In addition, MR and MB retained more surface crop residue than SD.