The use of riparian buffer strips is a possible strategy for controlling diffuse nitrate pollution of surface water in agricultural catchments. Data collected from paired buffered and unbuffered headwater catchments at three sites with conditions representative of much of the agricultural land in England and Wales, showed that grassed buffers did not substantially reduce nitrate–nitrogen concentrations entering the streams. Median nitrate–nitrogen levels observed in buffered catchments ranged from 7.6 to 18.8 mgN l−1, but peaked at up to 46.1 mgN l−1. The existence of preferential bypass flow paths during the winter flow events limited the effectiveness of nitrate removing processes within the strips. The findings suggest that grassed riparian buffer strips may not be effective in controlling diffuse nitrate pollution unless the hydrology of the strip allows a suitable environment for denitrification and/or plant uptake. Grassed buffer strips should be carefully targeted or, alternatively, engineered to ensure adequate residence time of solutes within the strip.