In order to quantify N-2-emissions from a spruce and a beech site at the Hoglwald Forest, a new measuring system was developed, that allowed simultaneous, direct determination of N-2- and N2O-emission with high accuracy (detection limit approx. 10 mug N m(-2) h(-1) for N-2 and <1 mug for N2O) using a gas-flow core method. This method requires exchange of the soil atmosphere with an artificial atmosphere, that differs only in that N-2 is substituted by He. The measuring system, the methodology of measurements and validation experiments are described in detail. Due to the huge heterogeneity of denitrification activity in different soil cores taken from our forest sites, no general trends of N-2 and N2O production in relation to soil moisture and temperature could be demonstrated. Based on reasonable number of measurements, this work gives for the first time an estimate of the magnitude of N-2-losses from temperate forest soils. Both the magnitude of N-2-emissions (spruce: 7.2+/-0.7 kg N-2-N ha(-1) yr(-1); beech: 12.4+/-3.1 kg N-2-N ha(-1) yr(-1)), as well as the N2O-N-2 ratio (spruce: 0.136+/-0.04; beech: 0.52+/-0.19) were significantly higher for soils from the beech sites as compared to soils from the spruce site. The results suggests that N-2-emissions from N-saturated forest soils, still receiving high loads of atmospheric N-deposition, are approx. 30% of atmospheric N-input at the spruce site, and approx. 50% at the beech site. Our results demonstrate that losses of nitrogen in the form of N-2 cannot be neglected in the context of calculating N-balances for given forest sites.