Denitrification is the most uncertain component of the nitrogen (N) cycle, hampering our ability to assess its contribution to reactive N (Nr) removal. This uncertainty emanates from the difficulty in measuring in situ soil N2 production and from the high spatiotemporal variability of the process itself. In situ denitrification was measured monthly between April 2013 and October 2014 in natural (organic and forest) and seminatural ecosystems (semi-improved and improved grasslands) in two UK catchments. Using the 15N-gas flux method with low additions of 15NO3 - tracer, a minimum detectable flux rate of 4 μg N m-2 h-1 and 0.2 ng N m-2 h-1 for N2 and N2O, respectively, was achieved. Denitrification rates were lower in organic and forest (8 and 10 kg N ha-1 y-1, respectively) than in semi-improved and improved grassland soils (13 and 25 kg N ha-1 y-1, respectively). The ratio of N2O/N2 + N2O was low and ranged from <1% to 7% across the sites. Variation in denitrification was driven by differences in soil respiration, nitrate, C:N ratio, bulk density, moisture, and pH across the sites. Overall, the contribution of denitrification to Nr removal in natural ecosystems was ∼50% of the annual atmospheric Nr deposition, making these ecosystems vulnerable to chronic N saturation.