Denitrification in subsoils may be a natural process able to decrease NO 3 contamination of groundwater. We studied this process and how it was affected by long-term fertilization practices. We collected soil to a depth of 2 m from four treatments in the Broadbalk Continuous Wheat Experiment. One treatment (FYM) had received manure applications annually since 1843, while the other treatments were N0 (0 kg N ha −1 since 1843), N4 (196 kg N ha −1 since 1968) and N6 (288 kg N ha −1 since 1984). Using the acetylene inhibition method, we measured N 2O production from anaerobic slurries of either unamended or amended (plus C and NO 3) samples of surface soil and three subsoil depth layers. These results will not equate to actual field denitrification rates, but the amended treatments (i.e. plus C and NO 3) will indicate the relative denitrification capacities of the soils. The unamended treatment will more accurately indicate the relative field denitrification rates. In the surface soils, denitrification rates were approximately proportional to total organic C (TOC) and they increased in the following order: N0, N4, N6, FYM. In the subsoil, denitrification declined with depth, and at 1.2 m was only a small fraction of that at the surface. There was no relationship between surface soil fertilization practices and either subsoil denitrification capacity, TOC, dissolved organic carbon (DOC) or total N. Indeed subsoil (60–200 cm) denitrification capacity was not significantly different between treatments. Denitrification capacity in unamended subsoils was not related to DOC. However, there was a significant relationship ( R 2=0.41) between TOC and denitrification rate. When amended soils were conditioned aerobically for 5 days, N 2O production increased, indicating the presence of a small community of denitrifiers which multiplied when given a C source. We conclude that subsoil denitrification has the potential to decrease NO 3 concentrations of percolating waters but is in reality limited by biologically available C. On our Broadbalk site, surface soil management had no effect upon the amount of `available' C in the subsoil.