Predicting the distribution of species has become a crucial issue in biodiversity research. Two kinds of model address this question: niche models, which are usually based on static approaches linking species distribution to habitat characteristics, and dispersal models, which are usually dynamic and process-based. We propose a model (NDM: niche and dispersal model) that considers the local presence of a species to result from a dynamic balance between extinction (based on the niche concept) and immigration (based on the dispersal concept), at a given moment in time, in a spatially explicit context. We show that NDM correctly predicts observed bird species and community distributions at different scales. NDM helps to reconcile the contrasting paradigms of metacommunity theory. It shows that sorting and mass effects are the factors determining bird species distribution. One of the most interesting features of NDM is its ability to predict well known properties of communities, such as decreasing species richness with decreasing patch size and increasing distance to the mainland, and the mid-domain effect at the regional scale, contrasting with predictions of much smaller effects at the local scale. NDM shows that habitat destruction in the matrix around patches of forest can affect the forest bird community, principally by decreasing the occurrence of typical matrix birds within the forest. This model could be used as the starting point for applied ecological studies on the management of species and community distributions.