The presence of ice shelves around 44% of the coastline of Antarctica means that the coastal sea ice is influenced by processes that take place at the base of ice shelves. In particular, theoretical studies estimate that basal meltwater may contribute as much as 0.2 m to the sea-ice thickness over significant portions of the sea-ice cover. Thus the sea ice in the vicinity of an ice shelf forms due to heat transfer to two heat sinks: the atmosphere and the ocean. The oceanic heat sink exists in the form of supercooled water as a result of interaction with an ice shelf. Ice crystals (called frazil or platelet ice) grow at depth in the water column or at the ice–water interface, these being most abundant when the ocean is at its coldest. We have investigated these processes in McMurdo Sound in the austral winters of 2003 and 2009. During each 8 month experiment, simultaneous observations were made of the growth of the landfast sea ice and the conditions in the ocean beneath. This paper is a preliminary overview of these findings. All field sites were located within 10 km of the ice shelf. However, the experiments in 2003 were conducted when McMurdo Sound was under the oceanographic influence of large icebergs, B15 and C19. By 2009 these icebergs had left the area and the Sound had relaxed towards its more normal condition. Consequently the waters in 2009 were generally warmer than in 2003. However by late June in both years much of the water column was below its surface freezing point, confirming extensive interaction with the ice shelf. From mid-winter a layer of water developed at the ice–water interface that was tens of mK below its in situ freezing point. This layer eventually became tens of meters thick and there was ample evidence of loose ice crystals being present in the water column. These disturb the columnar sea-ice growth, leading to changes in the ice structure. At the study sites, platelet ice that was incorporated into the first-year cover featured persistently after about 1 m growth, corresponding to late May to early June. This may be associated with an observed change in the surface circulation. Estimates of the contribution to the sea-ice cover by heat rejection to an oceanic sink show that it is significant and of the order of magnitude predicted by models.