In this paper, we develop a framework for (a) the study of sustainability of dynamic bioeconomic systems and (b) the definition of recovery paths from unsustainable situations. We assume that the system follows a sustainable trajectory if it evolves over time within a set of multidimensional constraints. We use the mathematical concept of viability to characterize sustainability. Recovery paths are studied with regards to their duration and their acceptability. This general framework is applied to the issue of recovering sustainable fisheries. We define sustainability in a fishery as the requirement that a set of economic, ecological, and social constraints is satisfied at all times. Recovery paths are characterized by the time required to obtain sustainable exploitation conditions in the fishery and by the acceptable recovery costs for fishermen. In particular, we identify the recovery path which minimizes the time of crisis under a minimum transition profit constraint. We then describe the trade-off between speed and accepted costs of recovery paths, by comparing "Hare"-like high-speed-high-cost strategies to "Tortoise"-like low-speed-low-cost strategies. We illustrate our results by means of a numerical analysis of the Bay of Biscay Nephrops fishery.