Under the United Nations Framework Convention on Climate Change (UNFCCC), industrialised countries and countries with economies in transition (the so-called Annex 1 countries) are encouraged to move towards more sophisticated approaches for national greenhouse gas reporting. Developing a model-based approach for estimating nitrous oxide (N2O) emissions from agricultural soils encompasses crucial steps of sound model selection, calibration, evaluation and upscaling of the model simulations to the regional level. To implement a model-based approach to simulate N2O emissions from agricultural soils in Switzerland, we selected the biogeochemical model DayCent considering its level of complexity and the availability of inputs required for regional simulations of N2O emissions. In the second step, we used extensive daily N2O flux observations from six cropland sites (four in Switzerland and two in France) and four grassland sites (two in Switzerland and two in Germany) to conduct automatic data-driven calibration of DayCent. After site-specific calibration, a leave-one-out (LOO) cross-evaluation was conducted for each land use type (i.e. cropland and permanent grassland) to assess the ability of the model to predict N2O emissions for sites it was not calibrated for. The LOO cross-evaluation resulted in an R2 of 0.63 for the prediction of N2O emissions from croplands and 0.65 from grasslands, compared to R2 values of 0.51 and 0.45 obtained with default parameterisation for croplands and grasslands, respectively. Overall, our results showed that the improvement in N2O predictions was usually associated with the adjustment of only a few parameters controlling the N cycle in soil (e.g., the maximum daily nitrification amount and the inflection point for the effect of water-filled pore space on denitrification). For grasslands, in addition to these parameters controlling N transformation in the soil, the adjustment of parameters related to N uptake by plants (thresholds of N sufficiency and deficiency and maximum biological N2 fixation) also affected the model’s ability to predict N2O emissions. These parameters also affected the simulation of N leaching, which is an indirect source of N2O. Overall, model-based estimates of N2O emissions were clearly closer to measurements than estimates based on commonly used emission factor (EF) approaches for both croplands and grasslands. Our results showed that, after data-driven calibration of only a few N cycle parameters in soil and plants, DayCent simulations are useful for reporting N2O emissions from agricultural soils in Switzerland. Therefore, in the last phase of the project, we performed a first upscaling of DayCent simulations of soil N2O emissions to the national level. For the upscaling, we gathered and processed geospatial data on of land use, soil properties, and weather variables at the country scale. In this preliminary model-based national simulation of N2O emissions, we stratified the territory and assumed an oversimplification of management practices. Nevertheless, the results of this project indicate that DayCent is an adequate model for reporting N2O emissions from agricultural soils in Switzerland. Obtaining more accurate management data is a crucial, necessary step towards establishing model-based estimates of N2O emissions for the national greenhouse gas inventory.