Measured data from two experimental sites in Canada were used to test the ability of the DeNitrification and DeComposition model (DNDC) to predict N2O emissions from agricultural soils. The two sites, one from eastern Canada, and one from western Canada, provided a variety of crops, management practices, soils, and climates for testing the model. At the site in eastern Canada, the magnitude of total seasonal N2O flux from the seven treatments was accurately predicted with a slight average over-prediction (ARE) of 3% and a coefficient of variation of 41%. Nitrous oxide emissions based on International Panel for Climate Change (IPCC) methodology had a relative error of 62% for the seven treatments. The DNDC estimates of total yearly emissions of N2O from the field site in western Canada showed an underestimation of 8% for the footslope landscape position and an overestimation of 46% for the shoulder position. The data input for the DNDC model were not of sufficient detail to characterize the moisture difference between the landscape positions. The estimates from IPCC guidelines showed an underestimation of 54% for the footslope and an overestimation of 161% for the shoulder. The results indicate that the DNDC model was more accurate than IPCC methodology at estimating N2O emissions at both sites. Key words: Nitrous oxide, DNDC, soil model, greenhouse gas, testing