Resilience-the ability of socio-ecological systems to withstand and recover from shocks-is a key research and policy focus. Definitions of resilience differ between disciplines, however, and the term remains inadequately operationalized. Resilience is the outcome of variable behavioral decisions, yet the process itself and the strategies behind it have rarely been addressed quantitatively. We present an agent-based model integrating four common risk management strategies, observed in past and present societies. Model outcomes under different environmental regimes, and in relation to key case studies, provide a mapping between the efficacy (success in harm prevention) and efficiency (cost of harm prevention) of different behavioral strategies. This formalization unravels the historical contingency of dynamic socio-natural processes in the context of crises. In discriminating between successful and failed risk management strategies deployed in the past-the emergent outcome of which is resilience-we are better placed to understand and to some degree predict their utility in the contemporary world.