The effects of grass strips on runoff interception, sediment trapping and soil loss were studied during two agricultural seasons. Field studies, conducted on loamy soil susceptible to sealing, allowed the comparison of three situations corresponding to buffer strip widths of 0, 3 and 6 m located at the downslope end of a winter wheat field. In 1997–98 the 6-m grass strips led to an average increase in infiltration of 87% with a coefficient of variation of 16% in comparison with a situation where no grass strip was present. The 3-m grass strip showed a slightly lower and more variable efficiency (average: 80%, coefficient of variation: 19% in 97/98). Maximum grass strip infiltrability was estimated at about 50 mm/h. Grass strips reduced the event mean sediment concentration by a factor of four on average. Sediments deposited in the grass strip were enriched in sand and coarse silt, whereas exported sediments contained twice as much clay and fine silt as the soil surface horizon. Net soil loss from the field was decreased by 76% in 96/97 and by 98% in 97/98 for the 6-m grass strip.