Several studies have used plant community biomass to predict species richness with varying success. In this study we examined the relationship between species richness and biomass for 36 marsh communities from two different watersheds. In addition, we measured several environmental variables and estimated the potential richness (the total number of species known to be able to occur in a community type) for each community. Above ground living and dead biomass combined was found to be weakly correlated with species richness (R²=0.02). Instead, a multiple regression model based on elevation (R²=0.47), salinity (R²=0.30), soil organic matter (R²=0.18), and biomass was able to explain 82% of the variance in species richness. It was found that environmental conditions could explain 89% of the variation in potential richness. Biomass had no relation to potential richness. When used as a predictor variable, potential richness was found to explain 72% of the variation in realized (observed) richness and biomass explained an addition 9% of the variance in realized richness. This finding suggests that realized richness in our system was controlled primarily by environmental regulation of potential richness and secondarily by biomass (as an indicator of competition). Further examination of the data revealed that when sites exposed to extreme environmental conditons were eliminated from the analysis, biomass became the primary predictor of realized richness and potential richness was of secondary importance. We conclude that community biomass has a limited capacity to predict species richness across a broad range of habitat conditions. Of particular importance is the inability of biomass to indicate the effect of environmental factors and evolutionary history on the potential species richness at a site.