Insight into the response of cereal yields to nitrogen fertilizer is fundamental to improving nutrient management and policies to sustain economic crop benefits and food sufficiency with minimum nitrogen pollution. Here we propose a new method to assess long-term (LT) regional sustainable nitrogen inputs. The core is a novel scaled response function between normalized yield and total net nitrogen input. The function was derived from 25 LT field trials for wheat, maize and barley in Europe, Asia and North America and is fitted by a second-order polynomial (R² = 0.82). Using response functions derived from common short-term field trials, with soil nitrogen not in steady state, gives the risks of soil nitrogen depletion or nitrogen pollution. The scaled LT curve implies that the total nitrogen input required to attain the maximum yield is independent of this maximum yield as postulated by Mitscherlich in 1924. This unique curve was incorporated into a simple economic model with valuation of externalities of nitrogen surplus as a function of regional per-capita gross domestic product. The resulting LT sustainable nitrogen inputs range from 150 to 200 kgN ha⁻¹ and this interval narrows with increasing yield potential and decreasing gross domestic product. The adoption of LT response curves and external costs in cereals may have important implications for policies and application ceilings for nitrogen use in regional and global agriculture and ultimately the global distribution of cereal production.