Various types of and modern Braun-Blanquet school viewpoints on homogeneity and heterogeneity of vegetation and their causes are discussed. A distinction is made between methodological minimum areas, MMA's (divided into qualitative and quantitative MMA's) and biological minimum areas, BMA's, divided into space, resistance and regeneration minima. Various definitions of the qualitative MMA are reviewed and some methods of analysis are discussed. It was found that the species-area curve of sample plots almost invariably follows the Fisher model, however with superimposed oscillations, which are regarded to represent a number of relative MMA's for each phytocoenosis (stand). These relative MMA's probably correspond to the elements of the compound mosaic pattern of which vegetation normally consists. A method to investigate the various biological minimum areas is expounded and results obtained in chalk grasslands, juniper scrub, oak woods and oligotrophic heath pools are briefly discussed. The species/area curve of BMA's appeared to follow either the Goodall or the Fisher or the Preston model. No saturation of the species number was obtained, not even in the largest stands examined (9 ha). Species numbers of stands were analysed in relation to size and age of the stands and to their degree of isolation. The value of the exponent z of the Preston formula appeared to be an unsuitable measure for the degree of isolation of these habitat islands. This z value was calculated for all species, for various taxonomic groups, including bryophytes and fungi, and for various dissemination types. There is no relation between size of stand, degree of isolation and number of dyschorous resp. eurychorous plants. Species numbers of phytocoenoses are obviously governed mainly by size, age and habitat quality (degree of disturbance).