Stimulation of nitrification and denitrification by long term (from years to decades) grazing has commonly been reported in different grassland ecosystems. However, grazing generally induces important changes in plant species composition, and whether changes in nitrification and denitrification are primarily due to changes in vegetation composition has never been tested. We compared soil nitrification- and denitrification-enzyme activities (NEA and DEA, respectively) between semi-natural grassland sites experiencing intensive (IG) and light (LG) grazing/mowing regimes for 13 years. Mean NEA and DEA (i.e. observed from random soil sampling) were higher in IG than LG sites. The NEA/DEA ratio was higher in IG than LG sites, indicating a higher stimulation of nitrification. Marked changes in plant species composition were observed in response to the grazing/mowing regime. In particular, the specific phytomass volume of Elymus repens was lower in IG than LG sites, whereas the specific volume of Lolium perenne was higher in IG than LG sites. In contrast, the specific volume of Holcus lanatus, Poa trivialis and Arrhenatherum elatius were not significantly different between treatments. Soils sampled beneath grass tussocks of the last three species exhibited higher DEA, NEA and NEA/DEA ratio in IG than LG sites. For a given grazing regime, plant species did not affect significantly soil DEA, NEA and NEA/DEA ratio. The modification of plant species composition is thus not the primary factor driving changes in nitrification and denitrification in semi-natural grassland ecosystems experiencing long term intensive grazing. Factors such as trampling, N returned in animal excreta, and/or modification of N uptake and C exudation by frequently defoliated plants could be responsible for the enhanced microbial activities.