The effects of ploughing or no-tillage of long-term grass and grass-clover swards on changes in organic C and N pools and on CO2 and denitrified gas emissions were investigated in a 3-year field experiment in 1996–99 near Penicuik, Scotland. The decrease in soil C content between 1996 and 1999 was 15·3 t/ha (95% confidence limits were 1·7–28·9 t/ha). Field estimates of CO2 losses from deep-ploughed, normal-ploughed and no-tillage plots were 3·1, 4·5 and 4·6 t/ha over the sampling periods (a total of 257 days) in 1996–98. The highest N2O fluxes were from the fertilized spring barley under no-tillage. Thus no-tillage did not reduce C emissions, caused higher N2O emissions, and required larger inputs of N fertilizer than ploughing. By contrast, deep ploughing led to smaller C and N2O emissions but had no effect on yields, suggesting that deep ploughing might be an appropriate means of conserving C and N when leys are ploughed in. Subsoil denitrification losses were estimated to be 10–16 kg N/ha per year by measurement of 15N emissions from incubated intact cores. A balance sheet of N inputs and outputs showed that net N mineralization over 3 years was lower from plots receiving N fertilizer than from plots receiving no fertilizer.