We illustrate the theoretical plausibility that fishery removal of sharks can indirectly alter predation pressure on different fish species via the behavioural responses of mesoconsumers released from predator intimidation. Our dynamic state variable model of foraging decisions by harbour seals, a mesopredator, predicted indirect effects of the removal of Pacific sleeper sharks on two species consumed by seals, Pacific herring and walleye pollock, as mediated by seal behaviour. Herring, a fatty fish, form near-surface aggregations that often are ephemeral and widely dispersed. Pollock are found in the deeper strata preferred by sharks and have lower energy density than herring, but also are larger and their more continuous distribution potentially makes them the more predictable resource for seals. During simulations, predation risk from sharks produced an asymmetric trophic cascade mediated by the seal's underutilisation of deeper, riskier strata. Risk management by seals reduced mortality on pollock, which required riskier access in deep strata, while increasing mortality on herring, which could be accessed with less risk in shallow strata. This effect, however, attenuated if herring were scarcer and seal energy state was poor. During shark removal scenarios, seals shifted to deeper strata, increasing pollock consumption and substantially decreasing use of herring; the proportional change was greatest if seals were in good energy state. Prior studies have addressed how shark declines might affect community structure through density responses by species consumed by sharks; earlier models incorporating antipredator behaviour of the mesopredator (i.e. Ecosim/Ecospace) allow for activity reduction but not for the spatial shifts that altered the asymmetric trophic cascade in our simulations. Our results suggest that shark declines might have stronger ecological consequences than previously recognised if we account for spatial and diet shifts by mesoconsumers released from shark intimidation.