While much is known about process level control on N2O production by nitrification and denitrification, knowledge of the environmental controls responsible for site variation in annual N2O fluxes on ecosystem level is low. Our goal was to improve existing concepts of controls on N2O fluxes. We measured N2O emission weekly or biweekly during 1 year in 11 temperate forest ecosystems using closed chambers. We identified three types of forest with different temporal emission patterns: forest with seasonal, event-based and background emission patterns. Comparison of annual data sets from literature showed that most temperate forests had low N2O emissions throughout the year (background emission pattern) with mean annual fluxes of 0.39¿0.27kgNha-1yr-1(n=21). Event-based emission patterns were observed during frost/thaw periods and after rewetting. Highest fluxes up to 72 kg N ha-1 were emitted from a drained alder forest with organic soil in 46 weeks, followed by well drained tropical and temperature forests with seasonal emission patterns and fluxes between 2--6(n=3) and 1--5kgNha-1yr-1(n=4), respectively. Seasonal emission patterns were explained by combined effect of high annual precipitations; broad leave trees; amount and structure of organic upper horizon; high mineral bulk densities; and plant community. These state variables reduce gas diffusivity so that oxygen demand by microorganism and roots exceeded oxygen supply during wet and warm periods (>10¿C). The resultant upper mean level was about 100&mgr;gN2O-Nm-2h-1 in both temperate and tropical forests. Annual N2O losses of the seasonal emission type were controlled by both duration and upper mean level of the periods with high emissions. We conclude that short-term controls of climate determine the duration of high emissions, whereas long-term controls by state variables determine the difference between background and seasonal emission types. ¿ 1999 American Geophysical Union