We present data from a comprehensive investigation carried out from 2007 to 2010, focussing on nitrogen pollution in the Oglio River basin (3800 km², Po Plain, Northern Italy). Nitrogen mass balances, computed for the whole basin with 2000 and 2008 data, suggest a large N surplus in this area, over 40 000 t N yr⁻¹, and increasing between 2000 and 2008. Calculations indicate a very large impact of animal husbandry and agricultural activities in this watershed, with livestock manure and synthetic fertilizers contributing 85% of total N inputs (about 100 000 t N yr⁻¹) and largely exceeding crop uptake and other N losses (about 60 000 t N yr⁻¹). Nitrogen from domestic and industrial origin is estimated as about 5800 and 7200 t N yr⁻¹, respectively, although these loads are overestimated, as denitrification in treatment plants is not considered; nonetheless, they represent a minor term of the N budget. Annual export of nitrogen from the basin, calculated from flow data and water chemistry at the mouth of the Oglio River, is estimated at 13 000 t N yr⁻¹, and represents a relatively small fraction of N inputs and surplus (∼12% and 34%, respectively). After considering N sinks in crop uptake, soil denitrification and volatilization, a large excess remains unaccounted (∼26 000 t N yr⁻¹) in unknown temporary or permanent N sinks. Nitrogen removal via denitrification was evaluated in the Oglio riverbed with stable isotope techniques (δ¹⁵N and δ¹⁸O in nitrate). The downstream final segment of the river displays an enriched nitrate stable isotope composition but calculations suggest a N removal corresponding to at most 20% of the unaccounted for N amount. Denitrification was also evaluated in riverine wetlands with the isotope pairing technique. Areal rates are elevated but overall N removal is low (about 1% of the missing N amount), due to small wetland surfaces and limited lateral connectivity. The secondary drainage channel network has a much higher potential for nitrogen removal via denitrification, due to its great linear development, estimated in over 12 500 km, and its capillary distribution in the watershed. In particular, we estimated a maximum N loss up to 8500 t N yr⁻¹, which represents up to 33% of the unaccounted for N amount in the basin. Overall, denitrification in surface aquatic habitats within this basin can be responsible for the permanent removal of about 12 000 t N yr⁻¹; but the fate of some 14 000 t remains unknown. Available data on nitrate concentration in wells suggest that in the central part of the watershed groundwater accumulates nitrogen. Simultaneously, we provide evidences that part of the stored nitrate can be substantially recycled via springs and can pollute surface waters via river-groundwater interactions. This probably explains the ten fold increase of nitrate concentration in a reach of the Oglio River where no point pollutions sources are present.