The effects of atmospheric nitrogen (N) deposition on ecosystem functioning largely depend on the retention of N in different ecosystem compartments, but accumulation and partitioning processes have rarely been quantified in long-term field experiments. In the present study we analysed for the first time decadal-scale flows and allocation patterns of N in a heathland ecosystem that has been subject to airborne N inputs over decades. Using a long-term 15N tracer experiment, we quantified N retention and flows to and between ecosystem compartments (above-ground/below-ground vascular biomass, moss layer, soil horizons, leachate). After 9 years, about 60% of the added 15N-tracer remained in the N cycle of the ecosystem. The moss layer proved to be a crucial link between incoming N and its allocation to different ecosystem compartments (in terms of a short-term capture, but long-term release function). However, about 50% of the 15N captured and released by the moss layer was not compensated for by a corresponding increase in recovery rates in any other compartment, probably due to denitrification losses from the moss layer in the case of water saturation after rain events. The O-horizon proved to be the most important long-term sink for added 15N, as reflected by an increase in recovery rates from 18 to 40% within 8 years. Less than 2.1% of 15N were recovered in the podzol-B-horizon, suggesting that only negligible amounts of N were withdrawn from the N cycle of the ecosystem. Moreover, 15N recovery was low in the dwarf shrub above-ground biomass (<3.9% after 9 years) and in the leachate (about 0.03% within 1 year), indicating still conservative N cycles of the ecosystem, even after decades of N inputs beyond critical load thresholds. The continuous accumulation of reactive forms of airborne N suggests that critical load-estimates need to account for cumulative effects of N additions into ecosystems.