Summary The concept of ecological communities play an important role in conservation planning and natural resource management. However, inherent uncertainties in the definition and identification of individual communities make it difficult to assess whether particular communities are present on particular sites and how they may be affected by proposed developments or management actions. If communities are poorly defined or misidentified, they may not perform their intended role as effective representations of biodiversity. We use a case study of forest communities dominated by Brown Barrel (Eucalyptus fastigata) in north-eastern New South Wales to demonstrate the value of quantitative floristic survey data for resolving robust and effective classifications of communities. Numerical analyses of an extensive set of floristic data suggested a re-configuration of a prior classification based largely on subjective interpretation. Although the general existence of assemblages dominated by Brown Barrel was confirmed, the new classification replaced three prior units with two assemblages that were more robust and better reflected the overall patterns in species composition. As only one of the two assemblages potentially warranted threatened status, the new classification allows scarce conservation resources to be targeted where they are most needed. The quantitative survey data also enabled a more detailed floristic description of the assemblages and provided a basis for maps of point locations and modelled habitat. These maps identified previously undocumented occurrences of the communities and helped to assess their extent of decline since European settlement. Improving the coverage of quadrat-based floristic sampling is therefore a valuable and cost-effective investment to inform better management of native vegetation and biodiversity.