Animal husbandry in Europe that sustained once wide-spread semi-natural grasslands has been replaced by maximum sustained yield agriculture and forestry. This transformation coincides with declining populations of species dependent on semi-natural grasslands. A key task is therefore to define benchmarks for landscape restoration in terms of well-planned functional habitat networks, i.e., green infrastructure. Using a representative example of the European landscape gradient between agricultural and forest landscapes in southern Sweden as a case study, we analyzed the historic range of variability of the total area, quality, and size of grassland patches, and compared this to the requirements of focal grassland species. Spatial data covering the past two centuries indicated a 75-80% loss of total grassland area. Three factors affected the functionality of grasslands as green infrastructure. First, during the period 1927-1976, the loss of all grassland areas with high nature values was 41-59%. Second, as a measure of alteration, the number of semi-natural grassland types declined from 5 to 1. Third, to address habitat fragmentation, an analysis of changes in grassland patch size showed that patches sufficiently large to support local populations of complete focal grasslands species assemblages declined by 89-100%. The cumulative effect of loss, alteration, and fragmentation over the past two centuries indicates that the functionality of semi-natural grasslands has declined by at least 98%. However, this estimate does not consider land use changes before 1800, reduced connectivity, and altered biotic and abiotic processes in both semi-natural grasslands and the surrounding matrix. We stress the need to define the historic range of variability as a benchmark in relation to species' requirements to maintain semi-natural grasslands as green infrastructure. Finally, integrated land management and governance that support multi-functionality of grasslands is needed.