We examined the response of a salt marsh food web to increases in nutrients at 19 coastal sites in Georgia. Fertilization increased the nitrogen content of the two dominant plants, Spartina alterniflora and Juncus roemerianus, indicating that added nutrients were available to and taken up by both species. Fertilization increased Spartina cover, height, and biomass and Juncus height, but led to decreases in Juncus cover and biomass. Fertilization increased abundances of herbivores (grasshoppers) and herbivore damage, but had little effect on decomposers (fungi), and no effect on detritivores (snails). In the laboratory, herbivores and detritivores did not show a feeding preference for fertilized versus control plants of either species, nor did detritivores grow more rapidly on fertilized versus control plants, suggesting that changes in herbivore abundance in the field were driven more by plant size or appearance than by plant nutritional quality.Community patterns in control plots varied predictably among sites (i.e., 17 of 20 regression models examining variation in biological variables across sites were significant), but variation in the effects of fertilization across sites could not be easily predicted (i.e., only 6 of 20 models were significant). Natural variation among sites was typically similar or greater than impacts of fertilization when both were assessed using the coefficient of variation. Overall, these results suggest that eutrophication of salt marshes is likely to have stronger impacts on plants and herbivores than on decomposers and detritivores, and that impacts at any particular site might be hard to distinguish from natural variation among sites.