Abstract. The Bering Sea is a highly productive ecosystem, supporting a variety of fish, seabird, and marine mammal populations as well as large commercial fisheries. Due to its unique shelf geometry and the presence of seasonal sea ice, the processes controlling productivity in the Bering Sea ecosystem span the pelagic water column, the benthic sea floor, and the sympagic sea ice environments. The BESTNPZ model has been developed to simulate the lower trophic level processes throughout this region. Here, we present a version of this lower trophic level model coupled to a three-dimensional regional ocean model for the Bering Sea. We quantify the model's ability to reproduce key physical features of biological importance as well as its skill in capturing the seasonal and interannual variations in primary and secondary productivity. We find that the ocean model demonstrates considerable skill in replicating observed horizontal and vertical patterns of water movement, mixing, and stratification, as well as the temperature and salinity signatures of various water masses throughout the Bering Sea. It is also able to capture the mean seasonal cycle of primary production observed on the data-rich eastern middle shelf. However, its ability to replicate domain-wide patterns in nutrient cycling, primary production, and zooplankton community composition, particularly with respect to the interannual variations that are important in a fisheries management context, remains limited.