Several studies have presented experimental evidence that diversity reduces invasibility in grassland communities. The interpretation of these results has been disputed recently and it was proposed that sampling effects were responsible for the observed decrease of invasibility with diversity. The experiments performed to date were not designed to adequately separate sampling from diversity effects. Using the establishment of native plant species in experimental plant communities as a model of invasibility, we show that the number of invaders decreased with increasing diversity. When the presence of particular species is included, their effects are dominant. Centaurea jacea showed a strong effect at low diversity, whereas Leucanthemum vulgare showed a very strong negative impact at each diversity level. The negative effect of the latter might be related to root-feeding nematodes that showed far higher abundance in plots with Leucanthemum. However, diversity remained a significant factor in determining the number of invading species and the numbers of an abundant invader.