Many attempts have been made to confirm or reject the unimodal relationship between disturbance and diversity stated by the intermediate disturbance hypothesis (IDH). However, the reasons why the predictions of the IDH apply or fail in particular systems are not always obvious. Here, we use a spatially explicit, individual-based community model that simulates species coexistence in a landscape subjected to disturbances to compare diversity-disturbance curves of communities with different coexistence mechanisms: neutrality, trade-off mechanism and intraspecific density dependence. We show that the shape of diversity-disturbance curves differs considerably depending on the type of coexistence mechanism assumed: (1) Neutral communities generally show decreasing diversity-disturbance curves with maximum diversity at zero disturbance rates contradicting the IDH, whereas trade-off communities generally show unimodal relationships confirming the IDH and (2) density-dependent mechanisms do increase the diversity of both neutral and trade-off communities. Finally, we discuss how these mechanisms determine diversity in disturbed landscapes.