<p>Water-level fluctuations drive plant species composition and abundance in freshwater wetlands. Receding levels stimulate seed-bank germination and increase plant diversity, while increasing levels drown flood-intolerant species and reduce diversity. Stable low water levels eventually lead to competitive exclusion and species reduction. Long-term studies of natural wetlands undergoing multiple water-level cycles are necessary to clarify plant responses. We examined a 50-year vegetation dataset from Cecil Bay on northern Lake Michigan that included the full range of water-level fluctuation observed since 1918. These data are well suited to test hypotheses about flood tolerance of coastal wetland species, rate of species response to changing water levels, and long-term flora and wetland stability. Our analyses reveal meadow-zone expansion and contraction with water-level change, strong correlation between plant density and richness with water levels, rapid diversity recovery following water-level drops, and characteristic species defining the meadow community. Water level and plant diversity are strongly negatively related, with a threshold at approximately the long-term average lake level. Cecil Bay data support the importance of seed banks in generating wet-meadow diversity, but data do not show annuals as a major feature of the marsh.</p>