In a study of an abandoned salt-marsh on Schiermonnikoog in the Netherlands, including an area where cattle grazing had been resumed, it was discovered that upper salt-marsh and dune species did not spread to the lower salt-marsh. Glass-house experiments, carried out on germination and dry matter production under increasing salinities, suggested that the high salinity of the lower salt-marsh was an important factor. On a grazed area of formerly abandoned salt-marsh, lower salt-marsh species spread to the upper salt-marsh and dune. This could be partly due to the better germination of lower salt-marsh species than upper salt-marsh and dune species, since the salinity of the uppermost cm was higher temporarily and locally in the grazed than in the abandoned salt-marsh. The canopy structure determining the amount of light reaching the soil also seemed to be an important factor enabling lower salt-marsh species to spread. In experiments, in which seeds of different taxa were sown into vegetation, lower salt-marsh species became established, particularly in the grazed area. The results of glass-house experiments, on germination and dry matter production under a range of light intensities, were consistent with studies of early establishment in upper salt-marsh and dune species in both the grazed and abandoned areas. Tidal seed dispersal resulted in the occurrence of lower salt-marsh species in the seed bank and seedling populations in the low and open canopy of the grazed upper salt-marsh and dune. Seed transport by cattle and geese could be important in the dispersal of some halophytes of the lower salt-marsh.