Although ecosystem-based fisheries management is often associated with trade-offs between conflicting demands for ecosystem services, the holistic ecological considerations the approach promotes may sometimes lead to novel solutions that benefit both conservation and fisheries. Directed fishing on large piscivorous fish can reduce predation on prey and thereby benefit those populations, but incidental take of prey species in these fisheries may negate or even reverse the benefit. Whether benefits from reduced predation outweigh the costs of increased fishing will depend on the relative strength of each mortality source and the sensitivity of the population to mortality at different life stages. In the California Current, predatory lingcod (Ophiodon elongatus) populations recovered rapidly from fishing exploitation over the past two decades, while recovery of some rockfish species (Sebastes spp.) has been slow, mainly because of low natural productivities. One management tool has been the adoption of rockfish conservation areas (RCAs) that prohibit bottom contact fishing gear. Because lingcod also inhabit RCAs, fishers have been unable to catch their sustainable lingcod quotas. Therefore, we explored the conditions under which opening RCAs to a selective lingcod fishery might permit rockfish recovery despite the potential bycatch. We developed a joint equilibrium model of the two populations and analysed scenarios to assess the sensitivity of the model's predictions to key uncertainties. The model suggests a wide range of fishery and ecological conditions under which a lingcod fishery may not harm rockfish populations. However, a sensitivity analysis indicated that the range of fishing scenarios where rockfish are not harmed is highly sensitive to assumptions regarding the nature of the trophic linkage between lingcod and rockfish. We conclude that consideration of trophic interactions may reveal new fishing opportunities that meet both ecological and human goals, but precise predictions of the outcomes will require more detailed models and adaptive management.