Gaseous nitrogen (N) losses from nitrification and denitrification (NO+N2O+N2) pathways contribute a significant fraction of the total N losses from cropland ecosystems. However, a large uncertainty exists in estimating the NO+N2O+N2 losses, hindering effective management of the global N budget. Here we proposed a novel isotope model, which considers N fertilizer, ammonia (NH3) volatilization and crop harvest after testing the steady state assumption of soil δ15N and N pool for croplands, and justified if it could be successfully applied to constrain NO+N2O+N2 losses from cropland ecosystems. We compiled the first bulk-soil δ15N dataset of 0-30cm soils (n=738) from croplands and produced a global map of cropland soil δ15N, which is crucial input data for N isotope model to quantify NO+N2O+N2 losses. The results show that the cropland soil δ15N ranges from 3.5 to 9.0‰, with a mean value of 6.6±0.8‰ (mean ± standard deviation). The estimated NO+N2O+N2 losses accounted for an average of 17±9% of N outputs and were 35.86±24.17 kg N ha-1 yr−1 in China’s rice paddies, with an increasing trend from Central China to South or North China. The estimations were comparable with the results from observation-constrained Denitrification-Decomposition modelling (38.9±4.8 kg N ha−1 yr-1) and in good agreement with experimental observations at site scale (R2=0.58). Our results suggest that soil N isotopes, as a quantitative tracer, provide a valuable alternative approach to constrain the NO+N2O+N2 losses in croplands at large geographic scales.