Hemiparasitic plants can reduce interspecific competition by parasitizing competitive species or by reducing the density of dominant plants. However, effects of hemiparasites have been suggested to be density dependent and to follow an optimum curve, with the highest diversity values occurring at intermediate hemiparasite densities. As a proof of principle, we tested whether this holds true for the two hemiparasitic species Melampyrum arvense and M. nemorosum. We conducted vegetation surveys in thermophilous fringe communities on the island of Saaremaa (Estonia) and tested how the local species richness of all vascular plants, and forbs and graminoids separately, vary along density gradi-ents of the two hemiparasitic species. In total, we investigated four spatially separated populations of M. arvense, and six separated populations of M. nemorosum. Within each population, we placed nine subplots along a Melampyrum density gradient. Using random slope, mixed-effects models, we found significant hump-shaped relationships of total vascular plant species richness with relative cover of M. arvense and M. nemorosum, with the highest richness values occurring at 13% and 40% cover, respectively. Starting at relatively low densities (> 29.7%) of M. arvense and relatively high densities (> 81.8%) of M. nemorosum, species richness was even lower than in plots without these hemiparasites. In contrast to results for total species richness, we found no significant effect of hemiparasite density when looking at graminoid and forb species separately. Our results help fill an important knowledge gap and underline the generality of density-dependent hemiparasite effects on local species richness. Our findings also clearly support the use of the density-gradient approach when the aim is to explore relationships between hemiparasites and species richness.