In Amazonia pastures are usually established after deforestation by sowing Urochloa grasses. Productivity declines due to nitrogen (N) limitation. Integrating N2- fixing legumes and grasses with Biological Nitrification Inhibition (BNI) capacity are considered as options for improving pasture sustainability. We studied grass-alone (GA) and grass–legume (GL) pastures on seven farms in Colombian Amazonia. Each treatment included grass with either low (Urochloa brizantha), intermediate (U. decumbens), or high (U. humidicola) BNI capacity. Biomass yields were measured over six harvests. Total N and δ15N of grasses and legumes and mineral N in topsoil (0–0.1 m) were analyzed in the first harvest. Results: GL pastures with high BNI U. humidicola showed a biomass yield benefit of 1.6 t ha−1 over six harvests and 9.3 kg N ha− 1 higher N yield for the first harvest compared to GA pastures. More than 70% of the legume N was derived from atmosphere, irrespective of Urochloa grass present. Plant δ15N indicated that Urochloa grasses obtained 20% less atmospheric N2 in GL than GA pastures. In contrast to U. brizantha, soil mineral N under U. humidicola did not increase significantly in GL pastures, indicating that BNI reduced nitrification and enhanced N use efficiency. Under conditions in Amazonia, legumes enhance the yield and N supply when associated with high BNI grass. The importance of different N sources and isotopic fractionation during N uptake of Urochloa grasses requires further investigation. These results highlight the potential of GL–high BNI systems to support pasture sustainability.