SUMMARYEmissions of nitrous oxide (N2O) and nitrogen gas (N2) from denitrification were measured using the acetylene inhibition method on drained and undrained clay soil during November 1980‐June 1981. Drainage limited denitrification to about 65% of losses from undrained soil. Emissions from the undrained soil were in the range 1 to 12 g N ha–1h–1while those from the drained soil ranged from 0.5 to 6 g N ha–1h–1giving estimated total losses (N2O + N2) of 14 and 9 kgN ha–1.Drainage also changed the fraction of nitrous oxide in the total denitrification product. During December, emissions from the drained soil (1.8±0.6 gN ha–1h–1) were composed entirely of nitrous oxide, but losses from the undrained soil (2.7 ± 1.1 g N ha–1h–1) were almost entirely in the form of nitrogen gas (the fraction of N2O in the total loss was 0.02). In February denitrification declined in colder conditions and the emission of nitrous oxide from drained soil declined relative to nitrogen gas so that the fraction of N2O was 0.03 on both drainage treatments. The delayed onset of N2O reduction in the drained soil was related to oxygen and nitrate concentrations. Fertilizer applications in the spring gave rise to maximum rates of emission (5–12g N ha–1h–1) with the balance shifting towards nitrous oxide production, so that the fraction of N2O w