Abstract The production of willow in short rotation is expected to result in low nitrate leaching losses as is the case with other permanent crops. However, there is a risk of leaching of nitrate during establishment when the plant cover is limited. Nitrate leaching was followed for three years from the establishment in 1993 of willow (Salix viminalis) at two sites, a coarse sand at Jyndevad and a loamy sand at Foulum. Two levels of nitrogen, 0 or 75 kg N/ha in NPK, were applied annually, though only 38 N/ha was applied in the first year. Leaching was calculated from usually weekly measurements of the nitrate concentration in soil solution and relating with the amount of percolation based on measured precipitation and changes in the soil water content over the sampling periods. Mean leaching for the three periods 1993–94, 1994–95 and 1995–96 amounted to respectively 142, 61 and 0 kg N/ha at Foulum and to 130, 9 and 4 kg N/ha at Jyndevad. The high leaching in the first year was caused by an unusually high content of mineral N in the soil at the start of the experiment. The absence of any leaching 1995–96 at Foulum was due to very low precipitation and a consequent lack of percolation. For the first period (1993–94) the fertilized treatment increased leaching by 32 kg N/ha compared to no fertilizer as a mean of the two sites, while the difference for the following two periods was reduced to 1–2 kg N/ha. It is concluded that application of nitrogen should be avoided in the year of planting of willows, while in the following years75 kg N/ha can be given without risk of increased leaching.