A quantitative measure of homogeneity, based on the average within‐group association between samples, is proposed as a multivariate measure of the information content of classifications and maps. Homogeneity analyses are used to investigate questions of scale and the choice of attributes in the context of vegetation mapping. Previous studies in classification and mapping of soil types suggested that an optimum map scale or number of classification groups can be defined using homogeneity. This does not seem to be the case with vegetation data although homogeneity analysis can be used to define the coarsest acceptable scale and to quantity the benefits of mapping at finer scales. Homogeneity analysis is used here to compare the information content of classifications derived from various attributes with one based on the whole flora. For the data set examined, a classification derived from canopy species composition is as informative as one based on full rloristic composition for the scales at which we would normally map, whereas an environmental classification is less so. Further applications of homogeneity analysis are suggested.