Balancing crop production and greenhouse gases (GHGs) emissions from agriculture soil requires a better understanding and quantification of crop GHGs emission intensity, a measure of GHG emissions per unit crop production. Here, we conduct a state‐of‐the‐art estimate of the spatial‐temporal variability of GHG emission intensities for wheat, maize, and rice in China from 1949‐2012 using an improved agricultural ecosystem model (DLEM‐AG2.0) and meta‐analysis covering 172 field‐GHG emission experiments. The results show that the GHG emission intensities of these croplands from 1949‐2012, on average, were 0.10‐1.31 kg CO₂‐eq kg^‐1, with a significant increase rate of 1.84‐3.58×10^‐3 kg CO₂‐eq kg^‐1 yr^‐1. Nitrogen fertilizer was the dominant factor contributing to the increase in GHG emission intensity in northern China and increased its impact in southern China in the 2000s. Increasing GHG emission intensity implies that excessive fertilizer failed to markedly stimulate crop yield increase in China but still exacerbated soil GHG emissions. This study found that overfertilization of more than 60% was mainly located in the winter wheat‐summer maize rotation systems in the North China Plain, the winter wheat‐rice rotation systems in the middle and lower reaches of the Yangtze River and southwest China, and most of the double rice systems in the South. Our simulations suggest that roughly a one‐third reduction in the current N fertilizer application level over these “overfertilization” regions would not significantly influence crop yield but decrease soil GHG emissions by 29.60%‐32.50% and GHG emission intensity by 0.13‐0.25 kg CO₂‐eq kg^‐1. This reduction is about 29% and 5% of total agricultural soil GHG emissions in China and the world, respectively. This study suggests that improving nitrogen use efficiency would be an effective strategy to mitigate greenhouse gas emissions and sustain China's food security.