Agroforestry represents an opportunity to reduce CO2 concentrations in the atmosphere by increasing carbon (C) stocks in agricultural lands. Agroforestry practices may also promote mineral N fertilization and the use of N2-fixing legumes that favor the emission of non-CO2 greenhouse gases (GHG) (N2O and CH4). The present study evaluates the net GHG balance in two adjacent coffee plantations, both highly fertilized (250 kg N ha?1 year?1): a monoculture (CM) and a culture shaded by the N2-fixing legume tree species Inga densiflora (CIn). C stocks, soil N2O emissions and CH4 uptakes were measured during the first cycle of both plantations. During a 3-year period (6-9 years after the establishment of the sys-tems), soil C in the upper 10 cm remained constant in the CIn plantation (+0.09 ± 0.58 Mg C ha?1 year?1) and decreased slightly but not significantly in the CM plantation (?0.43 ± 0.53 Mg C ha?1 year?1). Above-ground carbon stocks in the coffee monoculture and the agroforestry system amounted to 9.8 ± 0.4 and 25.2 ± 0.6 Mg C ha?1, respectively, at 7 years after establishment. C storage rate in the phytomass was more than twice as large in the CIn compared to the CM system (4.6 ± 0.1 and 2.0 ± 0.1 Mg C ha?1 year?1, respectively). Annual soil N2O emissions were 1.3 times larger in the CIn than in the CM plantation (5.8 ± 0.5 and 4.3 ± 0.3 kg N-N2O ha?1 year?1, respectively). The net GHG balance at the soil scale calcu-lated from the changes in soil C stocks and N2O emissions, expressed in CO2 equivalent, was negative in both coffee plantations indicating that the soil was a net source of GHG. Nevertheless this bal-ance was in favor of the agroforestry system. The net GHG balance at the plantation scale, which includes additionally C storage in the phytomass, was positive and about 4 times larger in the CIn (14.59 ± 2.20 Mg CO2 eq ha?1 year?1) than in the CM plantation (3.83 ± 1.98 Mg CO2 eq ha?1 year?1). Thus converting the coffee monoculture to the coffee agroforestr