Spatially explicit cropland greenhouse gas emission data are essential for identifying emission hotspots and guiding sustainable mitigation strategies. Here we develop high-resolution (5 arcmin) global maps of cropland emissions across 46 crop classes in 2020 by integrating sectoral datasets on synthetic fertilizer, manure, crop residue, in-field burning, rice cultivation and cultivated drained peatlands. Global croplands emitted 2.5 (95% CI 2.4–2.7) GtCO2e yr−1, with drained peatlands (35%), rice paddies (35%) and synthetic fertilizer (23%) as the primary contributors. Four crops—rice, maize, oil palm and wheat—accounted for 67% of total emissions. Emission areal intensities averaged 2 MgCO2e ha−1 globally, with higher intensities in Asia and Europe, where croplands also achieved high caloric productivity. Spatial correlations between emission intensity and production efficiency reveal geographic trade-offs between mitigation potential and food production. The resulting dataset establishes a unified global framework for a spatially explicit assessment of agricultural emissions and efficiency.