The phenomenon of bipolarity, one of the major disjunct distribution patterns on the face of the earth, has been investigated repeatedly since the mid-nineteenth century. Running through the many hypotheses that have been put forward to account for its occurrence, it is possible to detect two persistent themes: it is usually interpreted within a dispersal framework, and it is generally believed to be of comparatively recent origin. To many authors, the phenomenon is intimately linked to the Plio-Pleistocene glaciations. Recent palaentological investigations have extablished that bipolarity can now be traced back to at least the Early Jurassic period (i.e. 200 m.y.a.). Here it is well marked in the Pliensbachian stage by a variety of pectinacean bivalve taxa. Further bivalves indicate probable Middle Jurassic examples, but the phenomenon is more clearly seen in the Late Jurassic, especially in the Tithonian stage. At this time, in-oceramid, buchiid and oxytomid bivalve occurrences at northern hemisphere localities such as arctic Canada, N.W. Europe, Siberia, N.E. USSR and Japan can be matched with those in southern South America, Antarctica and Australasia. A striking Early Cretaceous (Aptian-Albian) bipolar pattern for the oxytomid Aucellina may be complemented by several infaunal bivalves, brachiopods and at least one gastropod. There is strong circumstantial evidence to suggest that bipolar molluscs continued to develop through the Cenozoic era. Such is the level of generic and subfamilial differentiation within certain living forms as to suggest that they must be the product of a considerable evolutionary history. It is likely that present-day distribution patterns of prosobranch gastropod groups such as th whelks (Buccinidae), together with certain fissurellids, littorinids, naticids and turrids, can be related to a late Paleogene-early Neogene phase of bipolarity. Many amphitropical taxa, in both the marine and terrestrial realms, have probable late Neogene-Pleistocene origins. It is possible to set the Jurassic and Cretaceous examples of bipolarity within a largely vicariant framework based upon the disintegration of the Pangean supercontinent. In this way the widespread ranges of putative Triassic ancestors were disrupted by tectonic processes in low latitude regions, although it should be emphasized that major climatic and oceanographic changes were almost certainly involved too. Similarly, it is possible to view late Paleogene earl Neogene bipolarity as a vicariant event, but this time with climatic change identified as the single most important agent. Widespread or cosmopolitan distributions are held to have formed during global cool phases (such as the late Eocene-early Miocene) only to be disrupted by global warming (such as in the late early Miocene). It is even possible to view Plio-Pleistocene patterns as, at leat in part, the products of vicariant events caused by rapid temperature and sea level shifts. Clearly, there is an urgent need here for more critical taxonomic data to test these various hypotheses. Phylogenetic studies of groups such as the Mesozoic bivalve superfamily Monotoidea and the Cenozoic Buccinidae, in particular, should constitute future rigorous tests. In so doing, they should also provide much useful information on the relative roles of dispersal and vicariance in promoting global disjunction in marine faunas. Repeated formation of bipolar patterns through geological time may have had important implications for modes of speciation and phenomena such as the origin of taxonomic diversity gradients.