A key component of ecosystem-based fisheries management (EBFM) is explicit consideration of trade-offs. Ecosystem models can be used to quantify trade-offs and focus discussion around objectives. Given large structural uncertainties inherent in ecosystem models, however, comparative approaches are recommended to identify results that are robust to model formulation. We developed ecosystem models of the continental shelf and slope of New South Wales, Australia, using two ecosystem modelling frameworks, Atlantis and Ecopath with Ecosim. The models were calibrated to emulate large-scale changes in ecosystem structure between 1976 and 1996, as predicted by data from fishery-independent trawl surveys. Calibrated models were projected forward under a range of "optimal" fishing efforts designed to achieve economic, conservation or biodiversity objectives. While there were large differences in model predictions for individual species, the models gave very similar qualitative results when ranking fishing policies and describing trade-offs. Our results illustrate the importance of identifying fishery objectives before build-up of fleet capacity, and the need to consider trade-offs when simultaneously stating multiple ecosystem-level goals. Our finding that structurally-distinct ecosystem models can provide consistent qualitative advice highlights the capacity of ecosystem models for informing strategic management questions, even in the presence of considerable uncertainty in ecosystem-level processes.