The effectiveness of vegetative buffer strips (VBS) for reducing herbicide transport has not been well documented for runoff prone soils. A multi‐year plot‐scale study was conducted on an eroded claypan soil with the following objectives: (1) assess the effects of buffer width, vegetation, and season on runoff transport of atrazine (ATR), metolachlor (MET), and glyphosate; (2) develop VBS design criteria for herbicides; and (3) compare differences in soil quality among vegetation treatments. Rainfall simulation was used to create uniform antecedent soil water content and to generate runoff. Vegetation treatment and buffer width impacted herbicide loads much more than season. Grass treatments reduced herbicide loads by 19‐28% and sediment loads by 67% compared to the control. Grass treatments increased retention of dissolved‐phase herbicides by both infiltration and adsorption, but adsorption accounted for the greatest proportion of retained herbicide load. This latter finding indicated VBS can be effective on poorly drained soils or when the source to buffer area ratio is high. Grass treatments modestly improved surface soil quality 8‐13 years after establishment, with significant increases in organic C, total N, and ATR and MET sorption compared to continuously tilled control. Herbicide loads as a function of buffer width were well described by first‐order decay models which indicated VBS can provide significant load reductions under anticipated field conditions.