Spring time nitrous oxide (N₂O) emissions from an old beech (<I>Fagus sylvatica</I> L.) forest were measured with eddy covariance (EC) and chamber techniques. The aim was to obtain information on the spatial and temporal variability in N₂O emissions and link the emissions to soil environmental parameters. Mean N₂O fluxes over the five week measurement period were 5.6&plusmn;1.1, 10&plusmn;1 and 16&plusmn;11 &mu;g N m^&minus;2 h^&minus;1 from EC, automatic chamber and manual chambers, respectively. High temporal variability characterized the EC fluxes in the trunk-space. To reduce this variability, resulting mostly from random uncertainty due to measuring fluxes close to the detection limit, we averaged the fluxes over one day periods. The variability in the chamber measurements was much smaller and dominated by high small scale spatial variability. The highest emissions measured by the EC method occurred during the first week of May when the trees were leafing and the soil moisture content was at its highest. If chamber techniques are used to estimate ecosystem level N₂O emissions from forest soils, placement of the chambers should be considered carefully to cover the spatial variability in the soil N₂O emissions. The EC technique, applied in this study, is a promising alternative tool to measure ecosystem level N₂O fluxes in forest ecosystems. To our knowledge, this is the first study to demonstrate that the EC technique can be used to measure N₂O fluxes in the trunk-space of a forest.