This study investigates spatial-temporal trends in N2O emissions from coffee production systems in Costa Rica with a focus on the effects of nitrogen fertilisation, topography and soil type. This is done by combining (i) multi-year continuous dynamic chamber measurements from sites with different fertilisation levels, (ii) static chamber measurements taken along a typical sloping coffee field and (iii) measurements from a laboratory incubation experiment with nutrient addition to different soil types. In the field and in the laboratory, additions included standard NPK fertiliser, ammonium nitrate (NH4NO3) as well as potassium nitrate (KNO3). Soils in a laboratory experiment were incubated under both drained and flooded conditions. Continuous measurements from automatic chambers show that annual N2O fluxes were dominated by bursts over few weeks following N-fertilisation with peak emissions up to 60 g N-N2O ha-1 day-1. A two-month slope experiment with static chambers after KNO3-fertilisation with 90 kg N ha-1 showed N2O significant differences between the highest daily emission rates from the top and the bottom of the slope (134 ± 20 g N-N2O ha-1 and 336 ± 104 g N-N2O ha-1, respectively) which can be explained by NO3- transport downhill and flooded conditions favouring denitrification at the bottom of the slope. Incubation experiments indicate that denitrification is the main process controlling N2O emissions but also that nitrification can result in low N2O emission rates under drained conditions. It can be concluded that the reported N2O emissions from the coffee agroforestry systems are generally low, but may be underestimated, as both poorly drained depressions functioning as N2O hotspots as well as temporal N2O bursts need to be taken into account.