Buffer strips have been widely recognized as to promote infiltration, deposition and sorption of contaminants for protecting surface water against agricultural contamination. However, such strips do not intercept all contaminants, particularly soluble ones. Although preferential flow (PF) has been suggested as one factor among several decreasing the efficiency of buffer strips, the mechanisms involved are not well understood. This project examines buffer strip efficiency at intercepting solutes when subsurface PF occurs. Two soluble sorbed tracers, FD&C Blue #1 and rhodamine WT, were applied on an agricultural sandy loam soil to evaluate the ability of a naturally vegetated buffer strip to intercept soluble contaminants. Rhodamine was applied about 15 m from the creek, while the Blue was applied 15 m to 165 m from the creek. Tracer concentration was measured over a two-year period in both the creek and the buffer strip through soil and water samples. Although the tracers traveled via different pathways, they both quickly moved toward the creek, passing beneath the buffer strip through the soil matrix. Our results demonstrate that the risk of water contamination by soluble contaminants is high in such systems, even when a well-vegetated buffer strip is used. The design of buffer strips should be modified to account for underground bypass, either by using plants that have deep, fine roots that do not favour PF or by adding a filter extending deep underground that can be regularly changed.