For 3 years we followed the complete annual cycles of N2O emission rates with 2-hour resolution in spruce and beech plantations of the H¿glwald Forest, Bavaria, Germany, in order to gain detailed information about seasonal and interannual variations of N2O emissions. In addition, microbiological process studies were performed for identification of differences in N turnover rates in the soil of a spruce and a beech site and for estimation of the contribution of nitrification and denitrification to the actual N2O emission. Both pronounced seasonal and extreme interannual variations of N2O emissions were identified. During long-term frost periods, while the soil was frozen, and during soil thawing, extremely high N2O emissions occurred, contributing up to 73% to the total annual N2O loss. The enormous N2O releases during the long-term frost period were due to high microbial N turnover rates (tight coupling of ammonification, nitrification, denitrification) in small unfrozen water films of the frozen soil at high concentrations of easily degradable substrates derived from the enormous pool of dead microbial biomass produced during the long-term frost period. Liming of a spruce site resulted in a significant increase in ammonification, nitrification, and N2O emissions as compared with an untreated spruce control site. The beech control site exhibited 4--5 times higher N2O emissions than the spruce control site, indicating that forest type itself is an important modulator of N2O release from soil. At all sites, nitrification contributed ~70% to the N2O flux, whereas denitrification contributed markedly less (~30%). Therewas a significant positive correlation between amount of in situ N input by wet deposition and magnitude of in situ N2O emissions. At the beech site, 10% of the actual N input was released from the soil in form of N2O, whereas at the spruce site the fraction was 0.5%. N2O emission rates were positively correlated with net nitrification rates. The results demonstrate the need for long-term measurements over several years for more precise estimates of annual N2O losses from forest ecosystems. On the basis of our results we conclude that the importance of temperate and boreal forests for the global N2O source strength may have been significantly underestimated in the past and that these forests contribute most likely ≫1.0 Tg N2O N. ¿ 1999 American Geophysical Union