Grazing generally benefits grassland biodiversity as it prevents shrub and tree succession. However, too intense grazing may have negative effects for example many grassland insects. EU-subsidies for grazing of some habitats, aimed at promoting biodiversity, still require a relatively intense grazing, and could therefore have negative consequences for some species. We quantified how such grazing affects habitat quality for the marsh fritillary butterfly, and how this influence its colonization-extinction dynamics and persistence. Specifically, we studied a metapopulation on Gotland (Sweden), where the marsh fritillary occupies unfertilized calcareous grassland with a naturally slow succession. We quantified the difference in larvae autumn nests between grazed and ungrazed habitat, and used this difference to adjust the ‘effective area’ of 256 habitat patches in a 50 km2 landscape. We then parameterized a metapopulation model based on the occurrence pattern of the adult butterfly, and simulated future population development under different grazing regimes. The results showed that ungrazed habitat harbored 4.8 times more nests than grazed habitat. Reducing the ‘effective area’ of grazed patches accordingly increased the local extinction probability and decreased colonization. Grazing all suitable habitat reduced the occupancy by over 80%, while no grazing increased the occupancy by up to 40%, based on projections of future dynamics. Current grazing is clearly too intense, and EU-subsidies are here, thus, a conservation measure with negative consequences for a threatened butterfly. To prevent this, subsidies for grazing need to be more flexible and better adapted to the prevailing soil conditions and requirements of the target species.