Grass-legume associations may offer a way of improving the N efficiency of dairy farming, but may also have an adverse impact on the environment by increasing leaching losses. Nitrate leaching from four cropping sequences with different grassland frequency and management (long-term grazed, long-term cut, cereals followed by 1 and 2-year grazed leys) were investigated on a loamy sand in central Jutland for both unfertilized grass-clover (perennial ryegrass (Lolium perenne L.)/white clover (Trifolium repens L.)) and fertilized perennial ryegrass (300 kg N/ha) swards during 1997–2002. Furthermore, 1 year (2001) of N2 fixation in 1-, 2- and 8-year-old grass-clover pastures was determined. Nitrate leaching from grazed unfertilized grass-clover was always considerably lower than from grazed fertilized ryegrass. The effect of grassland age on nitrate leaching was insignificant in grass-clover but clear in grazed ryegrass, where levels increased dramatically with sward age. In production years 6–8, leaching from grass-clover was only 9–13% of the comparable losses from ryegrass. Under the cutting regime grass-clover showed a significant reduction in both yield and N-removal with increasing sward age, whereas for ryegrass these figures did not show any decreasing trend. N2 fixation was lower in 8-year-old swards compared with fully established 2-year-old swards as a consequence of lower dry matter production, lower clover content and a lower proportion of clover-N derived from the atmosphere. The results from the present study indicate that the higher leaching losses observed in fertilized grass compared with unfertilized grass-clover systems were caused by (1) a reduction in N2-fixation in grass-clover over time and (2) a reduction in dry matter production in grass-clover over time, lowering the grazing intensity and the recycling of grassland N via animal excreta.