Evaluation of N2O flux has been one of the most problematic topics in environmental biogeochemistry over the last 10--15 years. Early ideas that we should be able to use the large body of existing research on terrestrial N cycling to predict patterns of N2O flux at the ecosystem scale have been hard to prove due to extreme temporal and spatial variability in flux. The vast majority of the N2O flux measurement and modeling activity that has taken place has been process level and field scale, i.e., measurement, analysis and modeling of hourly and daily fluxes with chambers deployed in field plots. It has been very difficult to establish strong predictive relationships between these hourly and daily fluxes and field-scale parameters such as temperature, soil moisture, and soil inorganic N concentrations. In this study, we addressed the question of whether we can increase our predictive understanding of N2O fluxes by examining relationships between flux and environmental parameters at larger spatial and temporal scales, i.e., to explore relationships between annual rather than hourly or daily fluxes and ecosystem-scale variables such as plant community and soil type and annual climate rather than field-scale variables such as soil moisture and temperature. We addressed this question by examining existing data on annual fluxes from temperate forest, cropland, and rangeland ecosystems, analyzing both multiyear data sets from individual sites as well as cross-site comparison of single annual flux values from multiple sites. Results suggest that there are indeed coherent patterns in annual N2O flux at the ecosystem scale in forest, cropland, and rangeland ecosystems but that these patterns vary by region and only emerge with continuous (at least daily) flux measurements over multiple years. An ecosystem approach to evaluating N2O fluxes will be useful for regional and global modeling and for computation of national N2O flux inventories for regulatory purposes but only if measurement programs are comprehensive and continuous. ¿ 2000 American Geophysical Union