ABSTRACT The design of cost‐efficient control strategies for invasive species that are too widespread and abundant for complete eradication, at least in the short term, will benefit from a rigorous analysis of invasion patterns and associated effects on native biodiversity. In this paper, the case of the invasive North American tree Prunus serotina in Flanders (Belgium) is presented. Our main objectives were to determine the susceptibility of forest stands to invasion by P. serotina and the subsequent effects of invasion on the understorey community. We used the large database of the first Flemish Forest Inventory. Multiple logistic regressions indicated that P. serotina occurred more frequently in privately owned, younger forest on coarse‐textured, dry soils (podzols), and the combination of these factors allowed us to correctly predict presence/absence of P. serotina in 70% of the validation plots. However, locational variables proved to be important as well, indicating that the invasion process is still ongoing. Prediction of P. serotina densities by means of multiple linear regressions was less successful. Effects on the understorey richness were analysed by comparing the number of species and the mean Ellenberg values between pairs of plots, only differing by the presence of P. serotina in the shrub layer. A reduction of the understorey richness following invasion was only pronounced on the more moist soils, while compositional changes mainly occurred on drier soils. It is concluded that priority for control should be given to landscapes with a low fraction of invaded stands and to forest stands located on more moist soils. However, using its potential to threaten native biodiversity as an argument for control should be done with care as further research is needed whether the observed negative effects are due to a species (i.e. native vs. non‐native) or a density effect (high vs. low).