Grassland age is increasingly recognised to be an indicator for present-day biodiversity, e.g. plant species richness, and is also important for other landscape functions. We developed a methodological approach to systematically assess the spatial distribution of grassland age in marginal European landscapes. This approach - applied to the Lahn-Dill Highlands (1270 km(2)), a marginal landscape in Hesse, Germany - comprises three steps: (1) in a two-stage stratification process, we pre-stratified the study area according to recent land-cover patterns and their changes between 1955 and 1995 (stratification I) and classified grassland types by combining data on soil moisture, base-richness, and elevation (stratification II). From 50 grassland types, we randomly selected 1000 representative grassland patches. (2) We determined the age of these patches by means of aerial photograph interpretation of a chronosequence dating back to 1953 and classified each patch with respect to the age classes young (<18 years), mid-aged (18-47 years), and old (>47 years). (3) Based on this information, we calculated grassland type-specific probabilities for grassland patches to belong to the respective age classes. These probabilities were projected to districts by direct extrapolation. An exemplary validation of extrapolation results for two test areas was performed. The results revealed that 49% of the investigated patches were old grassland. The remaining patches were mid-aged (36%) or young grassland (15%). The extrapolation results indicated accordingly a predominance of old grassland at the district scale. Occurrences of mid-aged grassland were concentrated in districts with a pronounced land-cover change, whereas young grassland is apparently evenly distributed across the study area. Validation results suggest that our approach is suitable for a realistic estimation of grassland age in marginal European landscapes. The method may be applied in landscape models of various disciplines that rely on large-scale information on grassland age.