The conventional selectivity paradigm is briefly reviewed and its performance examined from an ecosystem perspective. It is stressed that the overall (cumulative) selectivity of the harvest process in an ecosystem is the result of nested selection by fishers and fisheries of: (i) habitats; (ii) species assemblages; (iii) populations and (iv) individuals. A range of ecosystem models predict a strong impact of concentrated fishing (selective fishing) on the ecosystem structure stability, resilience and productivity. There seem to be advantages (in both yield and maintenance of ecosystem structure and functioning) to distribute fishing pressure broadly across available species and ecosystem compartments. Balanced harvesting was therefore defined by the workshop as a strategy that distributes fishing pressure across the wider possible range of trophic levels, sizes and species, in proportion to their natural productivity, reducing fishing pressure where it is excessive. The few attempts to verify the impacts predicted by models in real ecosystems with empirical data had limited success, indicating that such demonstration might be a significant challenge. Data from African small-scale fisheries were presented as a possible example of the capacity of multiple fisheries targeting an extremely broad range of species and sizes to extract high yield with limited impact on ecosystem structure. There are also a number of examples of surprising consequences of selectivity regulations resulting in either operational changes in the fishery or to unexpected shifts in the ecosystem. Emerging research priorities and management implications are reviewed.