Considerable empirical evidence suggests that aquatic ecosystems tend to support greater herbivory and higher ratios of herbivore to producer biomass (H:A) than do terrestrial ecosystems. Two general explanations have been proposed for this pattern: (1) the nutritional quality of primary producers differs among systems, leading to differences in the strength of top-down control by herbivores, or (2) food chain length differs among systems, which alters the magnitude of herbivory via trophic cascades. Under the first hypothesis, H:A ratios should be positively related to the nutrient content in the producers and independent of net primary production. Under the second hypothesis, H:A ratios should be related to net primary production but independent of producer nutrient content. We tested these two hypotheses using a compilation of papers from the literature reporting the biomass of producers and herbivores, net primary production and producer nutrient content across a wide range of ecosystems. H:A ratios were strongly and positively correlated with producer nutrient content across ecosystem types. H:A ratios were also unrelated to net primary production across ecosystem types, although a significant negative correlation existed within ecosystem types. Collectively, these patterns suggest that stoichiometric quality of producers, not differences in food chain length, is the dominant factor limiting food chain production across ecosystems.