Knowledge on seed bank density and species composition is crucial for predicting the probability that target species will establish in the plant community on a restored site. A general overview of data available for plant species occurring in coastal plant communities showed that information on seed persistence is up to now very limited. The available data suggest that restoration of coastal plant communities cannot rely on the seed bank, except for annual species of salt marshes, and that the seed bank is to a large extent composed of species of nutrient rich habitats. This was confirmed by two case studies in dune slacks and salt marshes on the Belgian coast. Seed density in dune slacks was found to be relatively high, but the seed bank contained almost exclusively seeds of species of nutrient rich habitats, resulting in a very low similarity ratio between seed bank and vegetation. Germination from the seed bank would rather hamper the establishment of target species because competitive pressure imposed by fast growing species of nutrient rich habitats would increase. In salt marshes, the similarity between seed bank and vegetation was higher, because there is a higher contribution of typical salt marsh species in the seed bank, although not all target species are equally represented. To allow predictions of future species composition on restored sites, seed bank studies should be an essential part of each coastal restoration project.