Insect herbivory can strongly affect ecosystem processes, and its relationship with plant diversity is a central topic in biodiversity-functioning research. However, very little is known about this relationship from complex ecosystems dominated by long-lived individuals, such as forests, especially over gradients of high plant diversity.2. We analysed insect herbivory on saplings of 10 tree and shrub species across 27 forest stands differing in age and tree species richness in an extraordinarily diverse subtropical forest ecosystem in China. We tested whether plant species richness significantly influences folivory in these highly diverse forests or whether other factors play a more important role at such high levels of phytodiversity.3. Leaf damage was assessed on 58 297 leaves of 1284 saplings at the end of the rainy season in 2008, together with structural and abiotic stand characteristics.4. Species-specific mean damage of leaf area ranged from 3% to 16%. Herbivory increased with plant species richness even after accounting for potentially confounding effects of stand characteristics, of which stand age-related aspects most clearly covaried with herbivory. Intraspecific density dependence or other abiotic factors did not significantly influence overall herbivory across forest stands.5. Synthesis. The positive herbivory-plant diversity relationship indicates that effects related to hypotheses of resource concentration, according to which a reduction in damage by specialized herbivores might be expected as host plant concentration decreases with increasing plant diversity, do not seem to be major determinants for overall herbivory levels in our phytodiverse subtropical forest ecosystem. We discuss the potential role of host specificity of dominant herbivores, which are often expected to show a high degree of specialization in many (sub)tropical forests. In the forest system we studied, a much higher impact of polyphagous species than traditionally assumed might explain the observed patterns, as these species can profit from a broad dietary mix provided by high plant diversity. Further testing is needed to experimentally verify this assumption.