Nitrate leaching was measured over seven years of nitrogen (N) addition in a paired-catchment experiment in Alptal, central Switzerland (altitude: 1200 m, bulk N deposition: 12 kg ha−1a−1). Two forested catchments (1500 m2each) dominated by Picea abies) were delimited by trenches in the Gleysols. NH4NO3 was added to one of the catchments using sprinklers. During the first year, the N addition was labelled with 15N. Additionally, soil N transformations were studied in replicated plots. Pre-treatment NO3−-N leaching was 4 kg ha−1a−1from both catchments, and remained between 2.5 and 4.8 kg ha−1a−1in the control catchment. The first year of treatment induced an additional leaching of 3.1 kg ha−1, almost 90% of which was labelled with 15N, indicating that it did not cycle through the large N pools of the ecosystem (soil organic matter and plants). These losses partly correspond to NO3−from precipitation bypassing the soil due to preferential flow. During rain or snowmelt events, NO3−concentration peaks as the water table is rising, indicating flushing from the soil. Nitrification occurs temporarily along the water flow paths in the soil and can be the source of NO3−flushing. Its isotopic signature however, shows that this release mainly affects recently applied N, stored only between runoff events or up to a few weeks. At first, the ecosystem retained 90% of the added N (2/3 in the soil), but NO3−losses increased from 10 to 30% within 7 yr, indicating that the ecosystem became progressively N saturated.