BackgroundTraditional habitat knowledge is an understudied part of traditional knowledge. Though the number of studies increased world-wide in the last decade, this knowledge is still rarely studied in Europe. We document the habitat vocabulary used by Csángó people, and determine features they used to name and describe these categories.Study area and methodsCsángó people live in Gyimes (Carpathians, Romania). The area is dominated by coniferous forests, hay meadows and pastures. Animal husbandry is the main source of living. Data on the knowledge of habitat preference of 135 salient wild plant species were collected (2908 records, 44 interviewees). Data collected indoors were counterchecked during outdoor interviews and participatory field work.ResultsCsángós used a rich and sophisticated vocabulary to name and describe habitat categories. They distinguished altogether at least 142–148 habitat types, and named them by 242 habitat terms. We argue that the method applied and the questions asked (‘what kind of place does species X like?’) helped the often implicit knowledge of habitats to be verbalized more efficiently than usual in an interview. Habitat names were highly lexicalized and most of them were widely shared. The main features were biotic or abiotic, like land-use, dominant plant species, vegetation structure, successional stage, disturbance, soil characteristics, hydrological, and geomorphological features. Csángós often used indicator species (28, mainly herbaceous taxa) in describing habitats of species. To prevent reduction in the quantity and/or quality of hay, unnecessary disturbance of grasslands was avoided by the Csángós. This could explain the high number of habitats (35) distinguished dominantly by the type and severity of disturbance. Based on the spatial scale and topological inclusiveness of habitat categories we distinguished macro-, meso-, and microhabitats.ConclusionsCsángó habitat categories were not organized into a single hierarchy, and the partitioning was multidimensional. Multidimensional description of habitats, made the nuanced characterization of plant species’ habitats possible by providing innumerable possibilities to combine the most salient habitat features. We conclude that multidimensionality of landscape partitioning and the number of dimensions applied in a landscape seem to depend on the number of key habitat gradients in the given landscape.