Many theoretical and experimental studies suggest that synergistic interactions between resources and predators influence foraging decisions and their fitness consequences. This framework, however, has been ignored almost completely by hypotheses on causes of the population decline of Steller sea lions (SSLs) (Eumetopias jubatus) in western Alaska. By comparing predictions from a dynamic state variable model to empirical data on the behaviour of individuals instrumented with satellite-linked time-at-depth recorders, we develop and find preliminary support for the hypothesis that, during winter in Prince William Sound, juvenile SSLs (a) underutilise walleye pollock, a predictable resource in deep strata, due to predation risk from Pacific sleeper sharks, and (b) underutilise the potential energy bonanza of inshore aggregations of Pacific herring due to risk from either killer whales, larger conspecifics, or both. Further, under conditions of resource scarcity-induced by overfishing, long-term oceanographic cycles, or their combination-trade-offs between mortality risk and energy gain may influence demographic parameters. Accordingly, computer simulations illustrated the theoretical plausibility that a decline of Pacific herring in shallow strata would greatly increase the number of deep foraging dives, thereby increasing exposure to sleeper sharks and mortality rates. These results suggest that hypotheses on the decline of SSLs should consider synergistic effects of predators and resources on behaviour and mortality rates. Empirical support for our model, however, is limited and we outline tasks for empirical research that emerge from these limitations. More generally, in the context of today's conservation crises, our work illustrates that the greater the dearth of system-specific data, the greater the need to apply principles of behavioural ecology toward the understanding and management of large-scale marine systems.