Developing and implementing conservation of Southern Ocean ecosystems is an exercise in risk management. Addressing the synergistic effects of climate change and fishing begs three questions: First, can we improve projections of future ecological change based on information we have at present? The tools for projecting change are essentially models: population models (with “add-ons” to take climate and fishing into account), species-distribution models, and food-web (end-to-end) models. I suggest some ways to advance these modelling approaches. Second, given that information and understanding of Southern Ocean ecosystems will remain forever imperfect, how can management now reduce the risk of irreversible or cascading degradation? I highlight recent work on the conservation role and design of large Southern Ocean Marine Protected Areas, especially in relation to CCAMLR’s principles of conservation and decision rules for finfish species. Third, given very limited (and likely declining) fieldwork resources over the next decades, how do we prioritise and implement ecological observation across the Southern Ocean? Vastly improved information gathering is needed to detect changes, improve dynamical understanding, and monitor the effectiveness of management. I suggest that new autonomous (robotic) sampling tools are required to address the challenges of affordability, spatial-scale, seasonal-coverage and ecological complexity.