This study examined how long-term insect herbivory, sand accretion, and the interaction of these factors affect patterns of plant succession on sand dunes. Individuals of a common dune species, Salix cordata, were either exposed to insect herbivory (by a specialist flea beetle, Altica subplicata) or protected from insect herbivory with cages from 1990 to 1992. Plant community composition and sand levels were then assessed from 1993 to 1996, during which time natural sand accretion occurred and flea beetles were rare. The goal was to examine how plant species composition and successional patterns were affected by past herbivory on one plant species, sand accretion, and an interaction between past herbivory and sand accretion. Past herbivory on Salix cordata caused a significant decrease in the proportional representation of herbaceous monocots in the plant community, as well as changes in the abundance of four of the 10 most common plant species. Other grasses and Aster/Solidago had greater increases in abundance in plots with past herbivory, Populus tremuloides had lesser decreases in abundance in plots with past herbivory, and Potentilla anserina had greater increases in abundance in plots without past herbivory. Past herbivory did not significantly affect changes in species richness over the study. Over the entire study, herbaceous monocots, herbaceous dicots, and horsetails increased in absolute abundance, but woody plant abundance decreased. Sand accretion significantly decreased the changes in abundance of three categories of plants: herbaceous monocots, herbaceous dicots, and woody plants. Only for horsetails was there was a positive relationship between change in abundance and sand accretion. As sand levels increased, the percentage of herbaceous dicots increased, whereas the percentage of herbaceous monocots decreased. There was a greater decrease in species richness with increasing amounts of sand deposition. For eight of the 10 most common species, there was a negative relationship between sand level and change in abundance; only beach grasses responded positively to sand accretion. Plant height was a significant factor in explaining which species increased vs. decreased in abundance; species that increased during the study were significantly taller than species that decreased. Clonal and nonclonal species responded similarly to sand accretion. There were very few interactive effects of past herbivory and sand level. Thus, past herbivory on Salix cordata had direct effects on the future composition of the plant community, via changes in plant abundance, but did not appear to modify the response of the plant community to sand accretion.