Southern Ocean palaeocirculation is clearly related to the formation of a continental ice sheet on Antarctica and the opening of gateways between Antarctica and the Australian and South American continents. Palaeoenvironmental proxy records from Southern Ocean sediment cores suggest ice growth on Antarctica beginning by at least 40 million years (Ma) ago, and the opening of Tasmania–Antarctic and Drake Passages to deep-water flow around 34 and 31 ± 2 Ma, respectively. So, the Eocene/Oligocene transition appears to mark the initiation of the Antarctic Circumpolar Current and thus the onset of thermal isolation of Antarctica with a first major ice volume growth on East Antarctic. There is no evidence for a significant cooling of the deep ocean associated with this rapid (< 350 000 years) continental ice build-up. After a long phase with frequent ice sheets growing and decaying, in the middle Miocene at about 14 Ma, a re-establishment of an ice sheet on East Antarctica and the Pacific margin of West Antarctica was associated with an increased southern bottom water formation, and a slight cooling of the deep ocean, but with no permanent drop in atmospheric pCO2. During the late Pleistocene on orbital time scales a temporal correlation between changes in atmospheric pCO2 and proxy records of deep ocean temperatures, continental ice volume, sea ice extension, and deep-water nutrient contents is documented. I discuss hypotheses that call for a dominant control of glacial to interglacial atmospheric pCO2 variations by Southern Ocean circulation dynamics. Millennial to centennial climate variability is a global feature, but there is contrasting evidence from various palaeoclimate archives that indicate both interhemispheric synchrony and asynchrony. The role of the Southern Ocean, however, in triggering or modulating climate variability on these time scales only recently received some attention and is not yet adequately investigated.