In this study we compared three relatively simple process based models, developed for the national scale (INITIATOR2), European scale (MITERRA) and global scale (IMAGE), with respect to their predictions of N budgets in response to changes in livestock, land use and agricultural management based on the IPCC B2 baseline scenario for the period 2000-2030. Predictions focus on NH3, N2O and NOx emissions to the atmosphere, N leaching to groundwater and N runoff to surface water. At the national and European scale, the comparison is quite good. For the year 2000, the results of INITIATOR2 and MITERRA at the national scale (the Netherlands) are within 10% for the N inputs (except for N fixation) and within 25% for most of the N outputs (crop uptake, NH3 and N2O emissions and N leaching). Larger differences, up to 60% occur for NOx emissions and N runoff. Similarly, the results of IMAGE and MITERRA compare quite well at the European scale (all EU 27 countries). Total N inputs compare within 10%, although individual sources such as grazing and deposition deviate up to 30%, and most of the N outputs deviate by less than 30% (crop uptake, NH3 and N2O emission). Larger differences up to 100% occur for NOx emissions and the sum of leaching and runoff. The comparability of predictions is different in 2030 as compared to 2000, due to differences in model assumptions. For example, an efficiency increase in N use in the period 2000-2030 is assumed in INITIATOR2 but not in MITERRA. Consequently, the N input by animal manure for the whole of the Netherlands is quite comparable in 2000, while it deviates by approximately 20% in 2030. Inversely, the estimated total NH3, N2O and NOx emissions by INITIATOR2 are approximately 20-40% higher than the estimates by MITERRA in the year 2000, but emissions in the year 2030 are comparable for both models. As an efficiency increase in N use in the period 2000-2030 is also assumed in IMAGE, similar discrepancies occur at the European scale when comparing MITERRA with IMAGE. Unlike the national and European scale predictions, results show that the area exceeding critical N loads and the average level of N exceedance is largely affected by spatial resolution of the input data. In this study, this holds specifically for effects on ground water quality (N leaching leading to NO3 concentrations exceeding the limit of 50 mg NO3/l), and to a lesser extent for impacts on terrestrial biodiversity (N deposition levels exceeding critical N loads). In summary, results imply that spatial aggregation has a limited effect on most national and continental scale N inputs and emission estimates but a moderate to large effect on the exceedance of critical N loads and critical NO3 concentrations, respectively. Largest uncertainties occur in the emissions of NOx and N leaching and runoff, both at the national and European scale. Differences in model results are mainly due to differences in the use of basic data, such as animal numbers and crop yields and in excretion, emission, uptake and leaching factors, and to a lesser extent related to differences in model structure.