This study analyses the spatial and temporal variability of N2O emissions from the agricultural soils of Belgium. Annual N2O emission rates are estimated with two statistical models, MCROPS and MGRASS, which take account of the impact of changes in land use, climate, and nitrogen-fertilization rate. The models are used to simulate the temporal trend of N2O emissions between 1990 and 2050 for a 10′ latitude and longitude grid. The results are also aggregated to the regional and national scale to facilitate comparison with other studies and national inventories. Changes in climate and land use are derived from the quantitative scenarios developed by the ATEAM project based on the Intergovernmental Panel on Climate Change-Special Report on Emissions Scenarios (IPCC-SRES) storylines. The average N2O flux for Belgium was estimated to be 8.6 × 106 kg N2O-N yr−1 (STD = 2.1 × 106 kg N2O-N yr−1) for the period 1990–2000. Fluxes estimated for a single year (1996) give a reasonable agreement with published results at the national and regional scales for the same year. The scenario-based simulations of future N2O emissions show the strong influence of land-use change. The scenarios A1FI, B1 and B2 produce similar results between 2001 and 2050 with a national emission rate in 2050 of 11.9 × 106 kg N2O-N yr−1. The A2 scenario, however, is very sensitive to the reduction in agricultural land areas (−14% compared with the 1990 baseline), which results in a reduced emission rate in 2050 of 8.3 × 106 kg N2O-N yr−1. Neither the climatic change scenarios nor the reduction in nitrogen fertilization rate could explain these results leading to the conclusion that N2O emissions from Belgian agricultural soils will be more markedly affected by changes in agricultural land areas.