Using a Bayesian model fitting approach, we developed a multispecies statistical catch-at-age model to assess trade-offs between predatory demands and prey productivities, focusing on the Lake Michigan pelagic fish community. We assessed these trade-offs in terms of predation mortalities and productivities of alewife (Alosa pseudoharengus) and rainbow smelt (Osmerus mordax) and functional responses of salmonines. Our predation mortality estimates suggest that salmonine consumption has been a major driver of historical fluctuations in prey abundance, with sharp declines in alewife abundance in the 1980s and 2000s coinciding with estimated increases in predation mortalities. While Chinook salmon (Oncorhynchus tshawytscha) were food limited during periods of low alewife abundance, other salmonines appeared to maintain a (near) maximum per-predator consumption across all observed prey densities, suggesting that feedback mechanisms are unlikely to help maintain a balance between predator consumption and prey productivity in Lake Michigan. This study demonstrates that a multispecies modeling approach that combines stock assessment methods with explicit consideration of predator–prey interactions could provide the basis for tactical decision-making from a broader ecosystem perspective.