The traditional fisheries management approach involves scientists providing their best assessment of the status and productivity of a resource. They then use these results to recommend a control measure, such as a total allowable catch (TAC), based upon some harvest control law, which is usually associated with a biological reference point (e.g. F(0.1)). Superficially, the operational management procedure (OMP), or equally the management strategy evaluation (MSE), approach for providing TAC recommendations may appear identical, as this often also links the results from some form of assessment to a harvest control law. However, the key difference is that the OMP/MSE approach involves simulation testing of the whole process that gives rise to the TAC recommendation within an adaptive management framework. This testing includes checks that application of the control law adopted will not lead to major problems, even if key perceptions about the resource happen to be in error; in other words, explicit account is taken of scientific uncertainties, in the spirit of the precautionary approach. Furthermore, quantitative evaluations are provided of the levels of catch to be anticipated in the medium term, and how these trade off against levels of risk of unintended depletion of the resource, to provide managers with a readily interpretable basis to choose between different management options. However, the process involves some problems in defining risk, which have yet to be resolved. Examples where ecosystem considerations have been taken into account in extending this OMP/MSE approach beyond the single-species level can be divided conveniently into two broad categories, depending on whether they concentrate primarily on operational (e.g. by-catch) or biological (e.g. predator-prey) interactions between species, and examples are given of each. To date, actual practical applications of this approach are more readily found for cases of operational interactions, particularly in the area of marine mammal by-catch. For practical applications involving biological interactions, the key limiting factor thus far is the paucity of data to estimate the form and magnitude of predation and competition interactions, which precludes confident computation of the trade-offs between harvest policy options that differ in the extents to which they concentrate upon different species. Nevertheless there are approximate approaches for dealing with this problem. We recommend the use of such approaches, while recognizing their limitations, until the data needed to develop more reliable models of biological interactions become available.