Aim To test if tree species richness and forest structure drive spatial variation in avian communities along a tropical elevation gradient and to present information about the role of detailed forest parameters. Location A 2000-m long elevational gradient of tropical forest on Mt. Cameroon, west-central Africa. Taxon Birds and trees. Methods We performed bird censuses and vegetation mapping at the same plots across six forested sites at elevations of 350, 650, 1,100, 1,500, 1,850, and 2,200 m a.s.l., with 16 plots per elevation. We tested the effects of elevation, forest structure and tree diversity on the species richness, functional diversity and β-diversity of birds (Bray-Curtis dissimilarity). We used conditional inference trees based on random forests (RF) to investigate these relationships across all elevation sites as well as within elevations. Results Both tree and bird species richness declined monotonically with elevation. Vegetation structure correlated with elevation, and all vegetation attributes significantly differed among elevations. The RF explained 70% of the variance in avian species richness, with the most important predictors being elevation, proportion of dead trees, tree species richness and herb layer coverage. We found that elevation (and shrub layerE2) was a particularly important predictor of avian functional diversity. We identified no important predictor of bird species richness after standardization within elevations, and the proportion of dead trees was the sole important predictor of functional diversity. Within-elevation β-diversity in avian community composition was determined by the dissimilarity of the tree community and differences in leaf area index, solar radiation and spatial distance. The functional dissimilarity was best explained by leaf area index. Main conclusions Apart from elevation itself, spatial distance even within elevations correlated with compositional and functional variation among avian assemblages. Forest structural traits can have a significant influence on distribution of birds. Thus, gaps in the spatial distribution of species such as along elevations might be caused by fine-scale recognition of suitable habitats.