Rivers are an important global sink for excess bioavailable nitrogen: they convert approximately 40% of terrestrial N-runoff per year (~47 Tg) to biologically unavailable N2 gas and return it to the atmosphere.1 Currently, riverine N2 production is conceptualised and modelled as denitrification.2-4 The contribution of anaerobic ammonium oxidation (or anammox), an alternate pathway of N2 production important in marine environments, is not well understood.5,6 Here we use in situ and laboratory measurements of anammox activity using 15N tracers and molecular analyses of microbial communities to evaluate anammox in clay, sand, and chalk-dominated river beds in the Hampshire Avon catchment, UK during summer, 2013. Anammox hzo gene abundance varied across the contrasting geologies. Anammox rates were similar across geologies but contributed different proportions of N2 production because of variation in denitrification rates. In spite of requiring anoxic conditions, anammox, most likely coupled to partial nitrification, contributed up to 58% of in situ N2 production in oxic, permeable riverbeds. In contrast, denitrification dominated in low permeability clay-bed rivers, where anammox contributes roughly 7% to the production of N2 gas. We conclude that anammox can represent an important nitrogen loss pathway in permeable river sediments.