The high concentration of the world's species in tropical forests endows these systems with particular importance for retaining global biodiversity, yet it also presents significant challenges for ecology and conservation science. The vast number of rare and yet to be discovered species restricts the applicability of species-level modelling for tropical forests, while the capacity of community classification approaches to identify priorities for conservation and management is also limited. Here we assessed the degree to which macroecological modelling can overcome shortfalls in our knowledge of biodiversity in tropical forests and help identify priority areas for their conservation and management. We used 527 plant community survey plots in the Australian Wet Tropics to generate models and predictions of species richness, compositional dissimilarity, and community composition for all the 4,313 vascular plant species recorded across the region (>1.3 million communities (grid cells)). We then applied these predictions to identify areas of tropical forest likely to contain the greatest concentration of species, rare species, endemic species and primitive angiosperm families. Synthesising these alternative attributes of diversity into a single index of conservation value, we identified two areas within the Australian wet tropics that should be a high priority for future conservation actions: the Atherton Tablelands and Daintree rainforest. Our findings demonstrate the value of macroecological modelling in identifying priority areas for conservation and management actions within highly diverse systems, such as tropical forests.