A test site was established in 1993 in the south-west of Sweden for studying a willow plantation as vegetation filter when irrigated with polluted agricultural drainage water. The soil was sand mixed with clay. The plantation was dense, 2x10(4) plants per hectare. Five clones of Salix viminalis and S. dasyclados were planted. The plantation was cut back after the first year of establishment. In spring 1994 furrows with a depth of 20 cm were made between each second plant row. Water was pumped into these furrows and distributed within the stands during the growing period. The irrigation water used was drainage water from covered pipes which drained 700 hectares of intensively fertilised agricultural land, The nitrogen content of the irrigation water was about 10-17 mg/l of nitrate nitrogen. Ground water pipes were installed in the stands from which samples were taken for the analyses of groundwater nitrogen content. The plantation was irrigated from May to November with about 11 mm per day. The mean evapotranspiration from the plantation was 5 mm per day, on average. The excess of water percolated through the soil and the nitrogen was taken up by the well-established root system. The amount of nitrogen delivered during 1995 to the stands with the irrigation water was 185 kg N per hectare. The nitrogen content in the leaves varied from 25 to 47 mg N/g DM during the growing period. The stem wood production of two-year-old stools for the five clones was 19-22 tonnes DM/hectare. The high nitrogen content in the leaves and the biomass production during the first rotation period indicated that the willows were well supplied with nitrogen and that the stands function as a vegetation filter, implying that the root system effectively took up and removed nitrogen from the irrigation water.