Environmental heterogeneity among sites can generate phenotypic and genetic variation facilitating differentiation and microevolution of plant populations. Badlands are desert‐like, predominantly vegetation‐poor habitats often embedded in (semi‐)dry grasslands. The desert‐like conditions of badlands demand extreme adaptation of plants, that is, phenotypic modifications in short‐term and/or natural adaptation in long‐term. However, detailed knowledge is missing about both plant phenotypic and genetic differentiation in this unique and widely occurring habitat type. The present study focused on the largest known badlands systems in Central Europe located in the “Drei Gleichen” region, a designated nature conservation area in Central Germany. Locations were suitable for this study in terms of having co‐occurring badlands and (semi‐)dry grassland habitats (sites) occupied by the pioneer plant Thymus praecox. Here, we studied the environmental preferences, morphological and functional trait variation, and genetic variation using microsatellite markers of T. praecox. Results revealed significant, mainly site‐dependent environmental, phenotypic, and genetic differentiation. In general, individuals in badlands are shorter in height and have lower patch sizes (length × width), relative growth rates, and smaller stomata. The PCA additionally unveiled slightly increased leaf robustness, trichome density, decreased stomatal conductance, fewer females, and earlier phenology in badlands. We interpret differentiation patterns as adaptive responses to light, temperature, drought, and nutrient stress conditions supported by reviewed literature. Genetic differentiation was strongest between local badlands and grassland sites, and clearly weaker among locations and between sites (in total) as indicated by GST, AMOVA, PCoA, and population structure. Our study supports the importance of small‐scale microhabitat conditions as a driver of microevolutionary processes, and the population's need for sufficient phenotypic variation and genetic resources to deal with environmental changes. We demonstrated that badlands are an appropriate model system for testing plant response to extreme habitats and that more research is needed on these fascinating landscapes.