A dynamic mass-balance model, ECOSIM, is used to compare the trophic impacts of harvesting small pelagic fish in three upwelling ecosystems, previously described using the ECOPATH mass-balance approach: Peru, Venezuela, and Monterey Bay. Four different simulated fishing regimes were considered for up to 100 years. Heavy exploitation of small pelagics resulted in increased biomass of their food and competitors, while their predators usually declined. Higher trophic levels were found to take the longest time to recover. Predicted ecosystem responses under assumptions of top-down and bottom-up control suggest that, in the Peruvian system, food limitation through bottom-up control cannot explain the switch of species from anchovy to sardine. Predictions of the fishing mortality for maximum sustainable yield in small pelagics were higher than those obtained by single species approaches. By asking "what if?" questions, this work highlights some of the insights that ECOSIM may offer in the development of ecosystem management principles for pelagic fisheries. Limitations and notes of caution in running the model are discussed.