The Eastern Bering Sea (EBS) and Gulf of Alaska (GOA) continental shelf ecosystems show some similar and some distinctive groundfish biomass dynamics. Given that similar species occupy these regions and fisheries management is also comparable, similarities might be expected, but to what can we attribute the differences? Different types of ecosystem structure and control (e.g. top-down, bottom-up, mixed) can imply different ecosystem dynamics and climate interactions. Further, the structural type identified for a given ecosystem may suggest optimal management for sustainable fishing. Here, we use information on the current system state derived from food web models of both the EBS and the GOA combined with dynamic ecosystem models incorporating uncertainty to classify each ecosystem by its structural type. We then suggest how this structure might be generally related to dynamics and predictability. We find that the EBS and GOA have fundamentally different food web structures both overall, and when viewed from the perspective of the same commercially and ecologically important species in each system, walleye pollock ( Gadus chalcogrammus ). EBS food web structure centers on a large mass of pollock, which appears to contribute to relative system stability and predictability. In contrast, GOA food web structure features high predator biomass, which contributes to a more dynamic, less predictable ecosystem. Mechanisms for climate influence on pollock production in the EBS are increasingly understood, while climate forcing mechanisms contributing to the potentially destabilizing high predator biomass in the GOA remain enigmatic. We present results of identical pollock fishing and climate-driven pollock recruitment simulations in the EBS and GOA which show different system responses, again with less predictable response in the GOA. Overall, our results suggest that identifying structural properties of fished food webs is as important for sustainable fisheries management as attempting to predict climate and fisheries effects within each ecosystem. [ABSTRACT FROM AUTHOR]