Understanding the patterns of biodiversity distribution and what influences them is a fundamental pre-requisite for effective conservation and sustainable utilisation of biodiversity. In this study we identified large-scale variation in tree species composition across tropical mountain forests in Africa (wet and moist, defined as > 800m asl). We used plot data from 23 sites in 12 countries in Africa (72,336 stems). As number of stems sampled per site was not even, we used Hill numbers. Elevation and climatic variables were extracted from BIOCLIM. We found relatively low alpha diversity (except in Nguti, Bakossi, Kahuzi, Itombwe and Udzungwa) but high beta diversity. A mean hierarchical classification, based Bray-Curtis distance (n = 10,000 replications), did not sperate geographical regions (west-central-east Africa), wetter from drier sites, nor large vs small mountains. Historical factors (forest refugia, not considered in this study) likely explain some of the patterns observed. Increased botanical exploration in some mountains could help better disentangle patterns observed.