The ecosystem approach to fisheries has been discussed since the 1980s. It aims to reduce risks from fisheries to whole, or components of, ecosystems, not just to target species. Precautionary approaches further aim to keep the risk of damage to a low level. Here, we provide a dynamic framework for spreading the ecosystems risk of fisheries in space and time, a method that can be used from the outset of developing fisheries and continually updated as new knowledge becomes available. Importantly, this method integrates qualitative and quantitative approaches to assess risk and provides mechanisms to both spread the risk, including enabling closed areas to help offset risk, and adjust catch limits to keep regional risk to a baseline level. Also, the framework does not require uniform data standards across a region but can incorporate spatially and temporally heterogeneous data and knowledge. The approach can be coupled with the conservation of biodiversity in marine protected areas, addressing potential overlap of fisheries with areas of high conservation value. It accounts for spatial and temporal heterogeneity in ecosystems, including the different spatial and temporal scales at which organisms function. We develop the framework in the first section of the paper, including a simple illustration of its application. In the framework we include methods for using closed areas to offset risk or for conserving biodiversity of high conservation value. We also present methods that could be used to account for uncertainties in input data and knowledge. In the second section, we present a real-world illustration of the application of the framework to managing risks of food web effects of fishing for Antarctic krill in the Southern Ocean. Last, we comment on the wider application and development of the framework as information improves.