Deep (>0.8km depth) fracture water with residence time estimates on the order of several Ma from the Witwatersrand Basin, South Africa contains up to 40μM of NO 3 -, up to 50mM N 2 (90 times air saturation at surface) and 1 to 400μM NH 3/NH 4 +. To determine whether the oxidized N species were introduced by mining activity, by recharge of paleometeoric water, or by subsurface geochemical processes, we undertook N and O isotopic analyses of N species from fracture water, mining water, pore water, fluid inclusion leachate and whole rock cores.The NO 2 -, NO 3 - and NH 3/NH 4 + concentrations of the pore water and fluid inclusion leachate recovered from the low porosity quartzite, shale and metavolcanic units were 10 4 times that of the fracture water. The δ 15N-NO 3 - and δ 18O-NO 3 - of the pore water and fluid inclusion leachate, however, overlapped that of the fracture water with the δ 15N-NO 3 - ranging from 2 to 7‰ and the δ 18O-NO 3 - ranging from 20 to 50‰. The δ 15N-NO 3 - of the mining water ranged from 0 to 16‰ and its δ 18O-NO 3 - from 0 to 14‰ making the mining water NO 3 - isotopically distinct from that of the fracture, pore and fluid inclusion water. The δ 15N-N 2 of the fracture water and the δ 15N-N from the cores ranged from -5 to 10‰ and overlapped the δ 15N-NO 3 -. The δ 15N-NH 4 + of the fracture water and pore water NH 3/NH 4 + ranged from -15 to 4‰. Although the NO 3 - concentrations in the pore water and fluid inclusions were high, mass balance calculations indicate that NO 3 - accounts for ≤10% of the total rock N, whereas NH 3/NH 4 + trapped in fluid inclusions or NH 4 + present in phyllosilicates account for ≥90% of the total N.Based on these findings, the fluid inclusion NO 3 - appears to be the source of the pore water and fracture water NO 3 - rather than paleometeoric recharge or mining contamination. Irradiation experiments indicate that radiolytic oxidation of NH 3 to NO 3 - can explain the fluid inclusion NO 3 - concentrations and, perhaps, its isotopic composition, but only if the NO 3 - did not attain isotopic equilibrium with the hydrothermal fluid 2billion years ago. The δ 15N-N, δ 15N-N 2 and δ 15N-NH 4 + suggest that the reduction of N 2 to NH 4 + also must have occurred in the Witwatersrand Basin in order to explain the abundance of NH 4 + throughout the strata. Although the depleted NO 3 - concentrations in the fracture water relative to the pore water are consistent with microbial NO 3 - reduction, further analyses will be required to determine the relative importance of biological processes in the subsurface N cycle and whether a complete subsurface N cycle exists. © 2011 Elsevier B.V.