The relationship between nitrous oxide (N2O) flux and N availability in agricultural ecosystems is usually assumed to be linear, with the same proportion of nitrogen lost as N2O regardless of input level. We conducted a 3-year, high-resolution N fertilizer response study in southwest Michigan USA to test the hypothesis that N2O fluxes increase mainly in response to N additions that exceed crop N needs. We added urea ammonium nitrate or granular urea at nine levels (0–292 kg N ha−1) to four replicate plots of continuous maize. We measured N2O fluxes and available soil N biweekly following fertilization and grain yields at the end of the growing season. From 2001 to 2003 N2O fluxes were moderately low (ca. 20 g N2O-N ha−1 day−1) at levels of N addition to 101 kg N ha−1, where grain yields were maximized, after which fluxes more than doubled (to >50 g N2O-N ha−1 day−1). This threshold N2O response to N fertilization suggests that agricultural N2O fluxes could be reduced with no or little yield penalty by reducing N fertilizer inputs to levels that just satisfy crop needs.