Three soybean production systems coupled with conventional tillage (CT) or no-tillage (NT) were evaluated for metolachlor and metribuzin loss in water from runoff. Metolachlor and metribuzin concentrations were similar in runoff from CT or NT soybean following winter wheat, with stubble burned, and CT or NT monocrop soybean. However, metolachlor and metribuzin concentrations were higher in runoff water from the CT doublecrop soybean planted into winter wheat stubble than NT. In all 3 yr of the study, the majority of total losses occurred in the first three runoff events. Total loss of metolachlor in runoff from the doublecrop system with winter wheat stubble present was 0.6% in 1991 to 1.1% in 1990 from CT, and 0.2% in 1990 to 0.6% in 1991 from the NT. Similar results were observed from the CT and NT doublecrop with burned winter wheat stubble and the monocrop systems. Metribuzin loss patterns were similar to metolachlor, but the fraction lost was slightly higher. Multi-variate regression was used to describe and predict herbicide losses, and R2values ranged from 0.66 to 0.98 for both metolachlor and metribuzin.