Agricultural soils are the main anthropogenic source of nitrous oxide (N2O), largely because of nitrogen (N) fertilizer use. Commonly, N2O emissions are expressed as a function of N application rate. This suggests that smaller fertilizer applications always lead to smaller N2O emissions. Here we argue that, because of global demand for agricultural products, agronomic conditions should be included when assessing N2O emissions. Expressing N2O emissions in relation to crop productivity (expressed as above-ground N uptake: ‘yield-scaled N2O emissions') can express the N2O efficiency of a cropping system. We show how conventional relationships between N application rate, N uptake and N2O emissions can result in minimal yield-scaled N2O emissions at intermediate fertilizer-N rates. Key findings of a meta-analysis on yield-scaled N2O emissions by non-leguminous annual crops (19 independent studies and 147 data points) revealed that yield-scaled N2O emissions were smallest (8.4 g N2O-N kg−1N uptake) at application rates of approximately 180–190 kg N ha−1 and increased sharply after that (26.8 g N2O-N kg−1 N uptake at 301 kg N ha−1). If the above-ground N surplus was equal to or smaller than zero, yield-scaled N2O emissions remained stable and relatively small. At an N surplus of 90 kg N ha−1 yield-scaled emissions increased threefold. Furthermore, a negative relation between N use efficiency and yield-scaled N2O emissions was found. Therefore, we argue that agricultural management practices to reduce N2O emissions should focus on optimizing fertilizer-N use efficiency under median rates of N input, rather than on minimizing N application rates.