In Belgium, managed soils are responsible for more than 60 % of N2O emissions from agriculture. This thesis aimed at gaining insight on the influence of farming practices and meteorological conditions on N2O emissions and flux dynamics in crops and grassland in Belgium. We worked on three research focus points: (1) the suitability of eddy covariance to measure N2O fluxes, (2) the influence of farming practices on N2O exchanges and (3) the weight of N2O in the greenhouse gas budget of managed soils. Three measurement campaigns were setup in Southern Belgium. In 2015, two sets of automated dynamic closed chambers were used to monitor N2O exchanges in a maize crop with the aim of studying the long-term effect of contrasting tillage practices (conventional tillage vs. reduced tillage) in experimental plots. In 2016, at the Lonzée ICOS Station, N2O emissions were measured by eddy covariance in a sugar beet crop from fertilization to harvest. Finally, in 2018, at the Dorinne ICOS Station, we set up a paired-flux tower measurement campaign to assess the impact of pasture restoration vs. business-as-usual on a control parcel. After 8 years of contrasting tillage practices, N2O and CO2 emissions were respectively 10 times and twice larger in the parcel under reduced tillage than in the conventionally tilled parcel. Reduced tillage practices, by limiting the mixing of crop residues to the topsoil layer, favor microbial development and increase N and C availability for microorganisms. In the sugar beet crop, we observed a large N2O emission peak following fertilization. While this event was expected, a surprising result was the inhibition of this burst after the seed-bed preparation, which constituted a first-time observation. We hypothesized that the microbiome was disturbed by this shallow soil disturbance. This hypothesis would benefit from further investigations. We estimated that N2O emissions accounted for at least 20 % of the net greenhouse gas budget of the crop, highlighting the importance of including N2O when ...