Plant community patterns in space and time may be explained by the interactions between competing plant species. The presented study investigates this in a nutrient and species poor ecosystem. The study presents a methodology for inferring competitive interactions from yearly vegetation inventories and uses this to assess the outcome of competitive interactions and to predict community patterns and dynamics in a Northwest-European dry heathland. Inferred competitive interactions from five consecutive years of measurements in permanent vegetation frames at a single dry heathland site were used to predict the community dynamics of C. vulgaris and D. flexuosa. This was compared with the observed plant community structure at 198 Danish dry heathland sites. Interspecific competition will most likely lead to competitive exclusion of D. flexuosa at the observed temporal and spatial scale, but the possibility of founder controlled alternative stable states (one dominated by C. vulgaris, the other by D. flexuosa) cannot be dismissed. The observed competitive interactions can only yield a vegetation pattern similar to the pattern observed across Danish dry heathland sites when disturbances are followed by initial dominance of D. flexuosa. The observed inter-annual variability in climatic conditions (summer precipitation and soil water content) affected both species in a similar way and did not substantially alter the competitive balance between the species. Our study supports the “successional mosaic hypothesis” where the maintenance of coexistence between C. vulgaris and D. flexuosa requires a storage effect and disturbance events followed by initial colonisation by D. flexuosa.