In dairy systems, grass–clover swards are fertilized with on–farm cattle slurry as a basic dressing and often supplemented with mineral fertilizer throughout the season. Uncertainty remains on consequences for N input from atmospheric N2-fixation by grassland legumes. The aim of this study was to examine the N response of slurry versus mineral N fertilization on clover dynamics, the proportion of N derived from the atmosphere (%Ndfa) and the quantitative N2-fixation (qBNF) using the 15N isotope dilution method. This was done in an on farm setting at two sites in two years with increasing rates of N fertilizer (0–480 kg available N ha-1) either of mineral N only or combined with application of cattle slurry. The leys were at one site composed a 2–species mixture of white clover (Trifolium repens L.) and ryegrass (Lolium perenne L.) and at the other site of a 4–species mixture of white clover, red clover (Trifolium pretense L.), festulolium (Festulolium braunii) and ryegrass. Species dynamics were significantly affected by fertilizer rate in the 2–species sward and by fertilizer rate and type in the 4–species sward. Reduction of the clover proportion in response to fertilization was due to a clover yield decrease (2–species sward) or a grass yield increase (4–species sward). In the 4–species sward, in treatments with mineral N only %Ndfa decreased markedly with N rate, whereas the corresponding slurry + mineral N treatments did not. Thus, apart from fertilization with mineral N only at one site the general picture was a relatively high (>80%) %Ndfa over the season when the N level was below 200–300 kg available N ha-1. The qBNF of the harvested biomass in sward without fertilization reached 193–216 kg N ha-1 in the 2–species sward and 203–286 kg N ha-1 in the 4–species sward. Clover persisted at high fertilization rates, but qBNF was reduced to 16–79 kg N ha-1 and 58–163 kg N ha-1 in 2– and 4–species swards, respectively. Interestingly, in the 4–species sward amended with slurry + mineral N, qBNF was high and independent of fertilization rate. We conclude that response of clover dynamics to N fertilization differed with fertilizer type and level – slurry have lower effect whereas the effect of mineral N only on N2-fixation is less predictable. •N response of grass–clover leys was investigated as affected by mineral N or slurry.•Dry matter yield and grass fraction increased as expected with N level.•Leys at two sites showed two patterns of N2-fixation activity response to fertilizer type.•Slurry in combination with mineral N sustained high N2-fixation until 240 kg N ha-1.•Clover N2-fixation response to mineral N only was less predictable among sites.