International initiatives such as the United Nations Framework Convention on Climate Change and the Kyoto Protocol require that countries calculate national inventolries of their greenhouse gas emissions. The objective of the present study was to develop and use a Tier II methodology for the inventory of N₂O emissions from agricultural soils in Canada.  Regional fertilizer-induced emission factors (FIEF) were first determined using available field experimental data.  Values for FIEF ranged from 0.0016 kg N₂O-N  kg^-1 N in the arid brown soil zone of the Prairies to 0.012 kg N₂O-N  kg^-1 N in the humid eastern provinces. A function relating FIEFs to climate (P/PE; P: precipitations PE: potential evapotranspiration) was developed and used to estimate annual emissions at the ecodistric level (avg. area: 430, 000 ha).  Algorithms were also developed to account for the effect of soil texture, spring thaw, topography, no-till, summerfallow and irrigation on N₂O emissions. Total direct N₂O emissions from agricultural soils averaged 36.8 Gg N yr^-1 between 1990 and 2004 with variations from 32.4 (2001) to 45.2 Gg N yr^-1 (2004).  These estimates are approximately 40% lower than estimates obtained using the IPCC Tier I methodology.  Differences in N₂O estimates between the two approaches mainly arise from the use of lower emission factors in the Tier II than in the Tier I methodology. Application of mineral N fertilizers was the most important single source of soil N₂O with average emissions during the 1990-2004 period of 13.5 Gg N yr^-1 or 36% of direct emissions.  Crop residues (9.0 Gg N yr^-1; 24%), grazing animals (7.6 Gg N yr^-1; 20%) and manure applied to soils (4.6 Gg N yr^-1; 12%) were the other major soil N₂O sources.