Abstract Early Devonian vertebrate faunal provinces are clearly defined for four regions: Euramerica, Siberia, China, and East Gondwana. The presence of osteostracans in southwest Siberia (Tuva, Minusa Basins) suggests either proximity of the Siberian block to Euramerica, or palaeogeographic separation of the Tuva region. The Knoydart fauna of Nova Scotia demonstrates that the Avalon Terrane was connected to Euramerica by Gedinnian time. Widespread antarctilamnid sharks in Gondwana suggest a distinctive Gondwana vertebrate fauna, isolated by marine barriers from Euramerica in the Early - Middle Devonian. Late Devonian patterns indicate faunal communication between Gondwana and Euramerica by Frasnian time, and between China and East Gondwana in the late Famennian. The Late Devonian base maps require anomalously wide latitudinal distributions for some taxa. Displacement of Turkey along the northern margin of Gondwana provides an intermediate occurrence of phyllolepid placoderms between disjunct distributions in Euramerica and East Gondwana, but the fossil data do not necessarily corroborate geological evidence for displacement. Biogeographic data generally must be interpreted in the context of palaeogeographic hypotheses, and lack of integration with geological and geophysical data sets has been a major problem. Hierarchical analysis using cladistic techniques has the potential for integrating biological, geological, and geophysical data, as illustrated in a cladistic analysis of the Williams and Hatcher model of Appalachian terranes. As an adjunct to map representation, an area cladogram enables a historical sequence of palaeogeographic events to be represented on a single diagram, together with crucial supporting evidence; it presents an analysis rather than synthesis of empirical data, and the hypothesis is more exposed to falsification.