Herbicides atrazine and metolachlor have been detected in water bodies across the world. The objective of this study was to assess the efficiency of 0-m, 3-m, 6-m and 9-m grass filter strips to reduce masses of dissolved metolachlor, atrazine and deethylatrazine (a degradation product of atrazine) exported in runoff. For that purpose, 16 uncultivated plots (3-m wide × 65-m long) with 0-m-, 3-m-, 6-m- and 9-m-long grass filter strips were setup in a completely randomized block design. During four seasons, masses of dissolved atrazine, metolachlor and deethylatrazine were determined for the first four to five rain events, under natural rain conditions, after atrazine and metolachlor application. Generally, grass filter strips reduced exported herbicide masses by more than 90% and influenced atrazine and metolachlor dissipation kinetics in the field. The 3-m grass filter strip (area ratio source/strip of 22:1) usually provided a reduction in exported herbicide masses similar to the 6- or 9-m grass filter strips. Therefore, under the present experimental soil and climate conditions, a grass filter strip of 3 m would be a good compromise between environmental protection of surface waters against atrazine and metolachlor contamination and conservation of agricultural land use. Such an approach contributes to the acceptability by producers to implement optimized best management practices such as vegetated filter strips for the preservation of the quality of water resources.