Agroforestry-based coffee production systems (AFs) contribute to climate change mitigation through carbon sequestration. However, it is unclear whether AFs produce lower nitrous oxide (N2O) and methane (CH4) emissions than the open-shade coffee production system. In addition, little to no evidence is available to explain the relationship between canopy cover levels and greenhouse gas (GHG) emissions in AFs. The aim of this study was to investigate N2O, CH4 and yield-scaled emissions in AFs with differing shade-tree canopy levels. Three canopy cover levels were identified: (i) dense shade (80 % canopy closure), (ii) medium shade (49 % canopy closure), and (iii) open-shade (full sun) production. To determine the effect of canopy cover on GHG emissions under varying soil fertility management practices, three soil fertilization strategies were included: (i) mineral fertilizer, (ii) compost, and (iii) control (i.e., without soil amendment). The results showed that N2O emissions were two-to-three times greater when there was dense canopy cover than from open-shade production. The effect of canopy cover on N2O emission was more pronounced under the mineral fertilizer treatment. CH4 emissions were 44–64 % greater under the open-shade production system than under AFs. The yield-scaled global warming potential of 1 kg of fresh coffee cherries was 0.72 kg CO2eq for open-shade production, 0.58 kg CO2eq for medium canopy cover and 0.52 kg CO2eq for dense canopy cover. This study provides the first evidence of the importance of considering canopy cover intensity when determining the spatial-temporal variations in GHG emissions from agroforestry systems.