Coho salmon (Oncorhynchus kisutch), Chinook salmon (Oncorhynchus tshawytscha) and steelhead (Oncorhynchus mykiss) in Puget Sound and the Strait of Georgia have exhibited declines in marine survival over the last 40 years. While the cause of these declines is unknown, multiple factors, acting cumulatively or synergistically, have likely contributed. To evaluate the potential contribution of a broad suite of drivers on salmon survival, we used qualitative network modelling (QNM). QNM is a conceptually based tool that uses networks with specified relationships between the variables. In a simulation framework, linkages are weighted and then the models are subjected to user-specified perturbations. Our network had 33 variables, including: environmental and oceanographic drivers (e.g., temperature and precipitation), primary production variables, food web components from zooplankton to predators and anthropogenic impacts (e.g., habitat loss and hatcheries). We included salmon traits (survival, abundance, residence time, fitness and size) as response variables. We invoked perturbations to each node and to suites of drivers and evaluated the responses of these variables. The model showed that anthropogenic impacts resulted in the strongest negative responses in salmon survival and abundance. Additionally, feedbacks through the food web were strong, beginning with primary production, suggesting that several food web variables may be important in mediating effects on salmon survival within the system. With this model, we were able to compare the relative influence of multiple drivers on salmon survival.