We estimated the long-term carbon balance [net biome production (NBP)] of European (EU-25) croplands and its component fluxes, over the last two decades. Net primary production (NPP) estimates, from different data sources ranged between 490 and 846 gC m−2 yr−1, and mostly reflect uncertainties in allocation, and in cropland area when using yield statistics. Inventories of soil C change over arable lands may be the most reliable source of information on NBP, but inventories lack full and harmonized coverage of EU-25. From a compilation of inventories we infer a mean loss of soil C amounting to 17 g m−2 yr−1. In addition, three process-based models, driven by historical climate and evolving agricultural technology, estimate a small sink of 15 g C m−2 yr−1 or a small source of 7.6 g C m−2 yr−1. Neither the soil C inventory data, nor the process model results support the previous European-scale NBP estimate by Janssens and colleagues of a large soil C loss of 90 ± 50 gC m−2 yr−1. Discrepancy between measured and modeled NBP is caused by erosion which is not inventoried, and the burning of harvest residues which is not modeled. When correcting the inventory NBP for the erosion flux, and the modeled NBP for agricultural fire losses, the discrepancy is reduced, and cropland NBP ranges between −8.3 ± 13 and −13 ± 33 g C m−2 yr−1 from the mean of the models and inventories, respectively. The mean nitrous oxide (N2O) flux estimates ranges between 32 and 37 g C Eq m−2 yr−1, which nearly doubles the CO2 losses. European croplands act as small CH4 sink of 3.3 g C Eq m−2 yr−1. Considering ecosystem CO2, N2O and CH4 fluxes provides for the net greenhouse gas balance a net source of 42–47 g C Eq m−2 yr−1. Intensifying agriculture in Eastern Europe to the same level Western Europe amounts is expected to result in a near doubling of the N2O emissions in Eastern Europe. N2O emissions will then become the main source of concern for the impact of European agriculture on climate.