A global inventory with 1°x1° resolution was compiled of emissions of nitrous oxide (N <sub>2</sub> O) to the atmosphere, including emissions from soils under natural vegetation, fertilized agricultural land, grasslands and animal excreta, biomass burning, forest clearing, oceans, fossil fuel and biofuel combustion, and industrial sources.A simple global model of the production potential of N <sub>2</sub> O in soils under natural vegetation was developed to analyze the relative importance of five major controls on N <sub>2</sub> O production: (i) input of organic matter, (ii) soil fertility, (iii) soil moisture status, (iv) temperature; and (v) soil oxygen status. Indices for the controls were derived from global gridded (1° x 1° resolution) data bases of soil type and texture, normalized difference vegetation index (NDVI) and climate. The model explains close to 60% of the variability found in measurements reported at about 30 sites in six different ecosystems throughout the world. The model results confirm conclusions from earlier studies that the major natural source regions of N <sub>2</sub> O are in the tropics.Literature data on N <sub>2</sub> O flux measurements from agricultural fields show that the fertilizerinduced N <sub>2</sub> O emission is higher for measurements covering longer periods than for measurements which represent short periods. A method to estimate the total annual direct N <sub>2</sub> O emission from fertilized fields was based on measurements covering periods of one year: N <sub>2</sub> O-N emission = I kg N ha <sup>-1</SUP>yr <sup>-1</SUP>plus 1.25 ± 1% of the amount of fertilizer N applied (kg N ha <sup>-1</SUP>yr <sup>-1</SUP>). This relationship was used to compile a global 1° x 1° resolution inventory of emissions of N <sub>2</sub> O from fertilized arable land.The inventory of N <sub>2</sub> O emission from animal excreta was based on estimates of nitrogen (N) excretion by various categories of domestic animals. The estimated global amount of N in animal excreta (-100 x 10 <sup>12</SUP>g N yr <sup>-1</SUP>) suggests that the associated N <sub>2</sub> O emission may be of the same order of magnitude as that caused by the use of synthetic N fertilizer (-80 x 10 <sup>12</SUP>g N yr <sup>-1</SUP>)To illustrate the difficulty to describe the cycling of N in ecosystems, N budgets were compiled for a deforestation sequence in the Atlantic Zone of Costa Rica. After forest clearing an important part of the soil organic N is mineralized. Part of the nitrate formed by nitrification of the mineralized N is lost via leaching, while most of the N loss occurs through denitrification. After a period of 3 to 5 years most of the easily decomposable material is lost, and denitrification and N <sub>2</sub> O fluxes decrease with time to levels lower than in the undisturbed forest. The global estimate of enhanced soil N <sub>2</sub> O emission from denitrification following tropical forest clearing, accounting for this decline of N <sub>2</sub> O fluxes along with ageing of the clearing, indicates that deforestation is an important global source.Global 10 x 1° resolution inventories were also compiled for N <sub>2</sub> O emissions from fossil fuel and fuelwood combustion, and industrial N2O sources. For N <sub>2</sub> O emissions from oceans and biomass burning, inventories from the literature were used. The complete inventory of annual N <sub>2</sub> O emissions including all sources, was compared with source estimates inferred from inversemodeling. The N <sub>2</sub> O inventories are in general agreement with inverse modeling results. However, there are major uncertainties, particularly in the tropics. The comparison between the emission inventory and the source estimates from inverse modelling, resulted in improved understanding of some sources:- The oceanic N <sub>2</sub> O emission may be higher than previous estimates; the 0°N-30°N latitudinal zone and the Antarctic ocean show much higher N <sub>2</sub> O fluxes than the mean global oceanic flux.- Most of the N <sub>2</sub> O from arable lands and grasslands including effects of synthetic fertilizers and animal excreta comes from the northern hemisphere. Inputs of N to soils from N deposition and from N fixation by leguminous crops are of the same order of magnitude as synthetic N fertilizer use. Their associated N <sub>2</sub> O release may also be similar in magnitude.- Fossil-fuel combustion and industrial N <sub>2</sub> O sources are dominant in the 30°N-90°N zone, while N <sub>2</sub> O from fuelwood combustion is mainly produced in the 0°N-30°N zone.- The major part of the N <sub>2</sub> O emitted from coastal marine and fresh water systems probably stems from the northern hemisphere.Global monthly estimates of N <sub>2</sub> O emissions were used to prescribe a three-dimensional atmospheric transport model. The simulated northern hemispheric N <sub>2</sub> O surface concentration was -1 ppb higher than in the southern hemisphere. This is in general agreement with atmospheric observations. The modeled N <sub>2</sub> O concentrations over strong source regions in continental interiors were up to 5 ppb higher than those over oceans. Predicted concentrations for the northern hemisphere were somewhat higher in summer than in winter, in agreement with the seasonality of N <sub>2</sub> O emissions. However, the atmospheric N <sub>2</sub> O concentration measurements show no seasonal variation in the northern hemisphere. The small seasonality in modeled atmospheric N <sub>2</sub> O concentrations for the southern hemisphere is more consistent with measurements. Inconsistencies between predicted and observed atmospheric N <sub>2</sub> O concentrations may be caused by overestimation of the seasonality in the northern hemisphere. The global N <sub>2</sub> O inventory used does not account for soil N <sub>2</sub> O consumption in temperate N-limited ecosystems and observed episodic emissions in temperate ecosystems during winter, early spring and autumn. These potential errors and possible underestimation of N <sub>2</sub> O emissions from combustion in winter may exaggerate the simulated seasonal trends.Another possible reason for the inconsistencies found between predicted and measured N <sub>2</sub> O concentrations is that seasonal trends in atmospheric N <sub>2</sub> O concentrations remain unobserved, because of the remote location of most monitoring stations. In addition, the precision of N <sub>2</sub> O measurements is not adequate for resolving seasonal trends.The major part of the atmospheric N <sub>2</sub> O increase stems from sources related to human food production. A growing world population will inevitably lead to more food demand. Moreover, an increasing portion of the synthetic fertilizers is used to increase animal production. At present, close to 40% of the global cereal production and 25% of the production of root and tuber crops is fed to animals. An important global reduction of N <sub>2</sub> O emission associated with food production could be achieved by a shift-away from animal production, by more efficient agricultural use of N, and by advanced fertilization techniques.Solutions to reduce N <sub>2</sub> O emission from fossil fuel combustion include technical options (e.g. development of new catalysts) and energy saving. Several industrial and chemical processes generate N <sub>2</sub> O, but so far only the N <sub>2</sub> O production from nitric and adipic acid production have been quantified. The global N <sub>2</sub> O emission from adipic acid production is currently decreasing as the major producers have agreed to reduce N <sub>2</sub> O emissions.