The Hudson Bay system is undergoing climate-driven changes in sea ice and freshwater inflow and has seen an increase in winter river inflow since the 1960s due in part to flow regulation for hydropower production. Southeast Hudson Bay and adjacent James Bay are at the forefront of these changes, with more than 1-month shortening of the season of sea ice cover as defined using satellite data, increases in winter inflow from the regulated La Grande River complex, and changes in coastal ice and polynya behavior described by Belcher Islands’ Inuit. In summer, there is a fresh coastal domain in southeast Hudson Bay fueled by river runoff and sea ice melt. To investigate winter oceanographic conditions and potential interactions between runoff and ice melt or brine in southeast Hudson Bay, we initiated the first winter study of the shallow waters surrounding the Belchers, collecting conductivity-temperature-depth (CTD) profiles and conductivity-temperature (CT) time series using under-ice moorings, and collecting water samples and ice cores during four campaigns between January 2014 and March 2015. Tandem measurements of salinity and δ18O were made for the water and ice samples to discriminate between freshwater sources (river runoff and sea ice melt). We find that southeast Hudson Bay, and particularly the nearshore domain southeast of the Belchers, is distinguished in winter by the presence of river water and strong surface stratification, which runs counter to expectations for a system in which local freshwater remains frozen on land until spring freshet (May–June) and sea ice growth is adding brine to surface waters. The amount of river water around the Belcher Islands increased significantly from fall through to late winter according to δ18O records of ice. The accumulation of river water in surface waters during the winter is directly associated with an accumulation of brine, which considerably exceeds the capacity of local ice formation to produce brine. We therefore infer that brine is advected into the study area together with river water, and that interplay between these properties establishes and maintains the level of surface stratification throughout winter. With reference to a NEMO ocean model simulation of winter circulation in the study area, we propose a conceptual model in which winter river inflow into James Bay drives the northward transport of both river water and brine captured near the surface, with reductions in brine-driven deep convection in the area’s flaw leads. While past changes in winter oceanographic conditions and sea ice cannot be reconstructed from the few available scientific data, the presence of significant runoff in winter in southeast Hudson Bay implies heightened sensitivity to delayed freeze-up under a warmer climate, which will have the effect of reducing brine early in the winter, also promoting increased stratification and river plume transport.