Effectiveness of vegetative buffer strips for herbicide retention from agricultural runoff was evaluated in a two-year natural rainfall study. A source area of 0.41 ha (mainly Canisteo silty clay loam soil), having an average slope of 3%, was fall chisel-plowed, spring disked, and planted to corn. Three herbicides (atrazine, metolachlor, and cyanazine) were applied to the source area in each spring. Six vegetative buffer strips, 1.52 m wide ¥ 20.12 m long, were isolated with metal borders downslope of the source area in a well established bromegrass (Bromus inermis) waterway. These strips provided for three replications of two drainage to buffer area ratio treatments of 15:1 and 30:1. Herbicide retention was dependent on the antecedent moisture conditions of the strips. These retentions ranged from 11 to 100% for atrazine, 16 to 100% for metolachlor, and 8 to 100% for cyanazine. Herbicide retention by the buffer strips for the two treatments were not significantly different for the observed storm events. Herbicide concentrations in solution in outflow from the strips were less than the inflow concentrations for all the three herbicides. Infiltration was the key process for herbicide retention by the buffer strips, although there was some adsorption to in-place soil and/or vegetation. Metolachlor concentrations in sediment increased in outflow for the two treatments; however, the opposite was true for atrazine and cyanazine. Herbicide retention by sediment deposition in the strip represented about 5% of the total herbicide retention by the buffer strips. The buffer strips were found to have high percent sediment retention, ranging from 40 to 100%; thus, the strips would be more effective for retaining strongly adsorbed herbicides.