Tropical forest soils are a significant source for N2O and NO. Current estimates of N2O and NO emissions are uncertain due to the limited number of field measurements and model input data. Furthermore, considerable spatial and temporal variability exists due to variation of soil properties, vegetation characteristics and meteorology. We used a process-based model (ForestDNDC-tropica) to estimate N2O and NO emissions from the entire (970 km2) tropical highland forest (Nyungwe) in southwestern Rwanda. Scaling these results to that regional level using legacy soil, meteorological and simulated vegetation data we found in most cases agreement between N2O and NO measurements and model predictions. Limited agreement was found for acid soils with high clay content and reduced metals, indicating that abiotic N2O and NO forming processes in acidic soils might be under-represented in the current ForestDNDC-tropica model. The Nyungwe forest was estimated to emit 439 t N2O-N year− 1 (2.8–5.5 kg N2O-N ha− 1 year− 1) and 244 t NO-N year− 1 (0.8–5.1 kg N ha− 1 year− 1), corroborating previous studies in tropical forests and highlighting that also tropical highland rainforest soils are a major source of atmospheric N2O and NO. The uncertainty for the N2O and NO emission estimates was 153 and 50 t N2O-N year− 1 and 36 and 16 t NO-N year− 1 considering uncertainty in model input data and annual variability, respectively. The results showed that soil bulk density and pH were the most influential factors driving spatial variation and model uncertainty. To improve global model-based estimates of N2O and NO emission from tropical forest focus should therefore also be oriented in delivering more detailed soil and vegetation data