In Madagascar, no-tillage practices were developed since the early 90s to prevent soil erosion and improve soil fertility. Although such practices have helped to restore soil carbon in most cases, the impact on N2O emissions has not been investigated yet. The soil N2O fluxes and concentrations were measured during the growing season of an intercropping maize-soybean on a clayey soil of the Malagasy Highlands. Management treatments consisted of direct seeding mulch based cropping system (DMC) and traditional hand-ploughing after the preceding crop residues were harvested (HP), both with low N inputs (55-57 kg N ha− 1). No significant difference in N2O emissions was observed between treatments (DMC vs. HP). The N2O fluxes were weakly correlated to soil mineral N contents (R2 = 0.13; P = 0.03) while no relationship was emphasized with soil water filled pore space (WFPS). N2O concentrations in the soil atmosphere were correlated to fluxes at the soil surface and to soil WFPS. N2O emissions at the soil surface were low ranging from 0 to 8.84 g N-N2O ha− 1 d− 1, probably due to the low mineral N content of soil. The cumulative annual N2O emission was 0.26 kg N ha− 1 for both systems. The corresponding N loss as N2O-N was around 0.5% of applied N. This is in the uncertainty range of IPCC N2O emission factor (EF), but the IPCC EF mean estimate (1%) would overestimate true N2O emissions for the soil under evaluation.