Losses of plant species diversity can affect ecosystem functioning, with decreased primary productivity being the most frequently reported effect in experimental plant assemblages, including tree plantations. Less is known about the role of biodiversity in natural ecosystems, including forests, despite their importance for global biogeochemical cycling and climate. In general, experimental manipulations of tree diversity will take decades to yield final results. To date, biodiversity effects in natural forests therefore have only been reported from sample surveys or meta-analyses with plots not initially selected for diversity. We studied biomass and growth of subtropical forests stands in southeastern China. Taking advantage of variation in species recruitment during secondary succession, we adopted a comparative study design selecting forest plots to span a gradient in species richness. We repeatedly censored the stem diameter of two tree size cohorts, comprising 93 species belonging to 57 genera and 33 families. Tree size and growth were analyzed in dependence of species richness, the functional diversity of growth-related traits, and phylogenetic diversity, using both general linear and structural equation modeling. Successional age covaried with diversity, but differently so in the two size cohorts. Plot-level stem basal area and growth were positively related with species richness, while growth was negatively related to successional age. The productivity increase in species-rich, functionally and phylogenetically diverse plots was driven by both larger mean sizes and larger numbers of trees. The biodiversity effects we report exceed those from experimental studies, sample surveys and meta-analyses, suggesting that subtropical tree diversity is an important driver of forest productivity and re-growth after disturbance that supports the provision of ecological services by these ecosystems.