Abstract The tectonic evolution of Archaean granite-greenstone terranes remains controversial. Here this subject is reviewed and illustrated with new data from the Slave craton. These data show that a thick, c. 2.7Ga, pillow basalt sequences extruded across extended sialic basement of the Slave craton at a scale comparable with that of modern large igneous provinces. The pillow basalts do not represent obducted oceanic allochthons. Basement-cover relationships argue for autochthonous to parautochthonous development of the basaltic greenstone belts of the west-central Slave craton, an interpretation that is further supported by geochemical and geochronological data. Similar data exist for several other cratons and granite-greenstone terrains, including the Abitibi greenstone belt of the Superior craton, where stratigraphic and subtle zircon inheritance data are equally incompatible with accretion of oceanic allochthons. Many classical granite-greenstone terrains, including most well-documented komatiite occurrences, thus appear to have formed in extensional environments within or on the margins of older continental crust. Closest modern analogues for such basalt-komatiite-rhyolite-dominated greenstone successions are rifts, marginal basins and volcanic rifted margins. Indeed, these environments have high preservation potential compared with fully oceanic settings. Collapse and structural telescoping of these highly extended volcano-sedimentary basins would allow for the complex structural development seen in granite-greenstone terrains while maintaining broadly autochthonous to parautochthonous tectonostratigraphic relationships. Seismic reflection profiles cannot discriminate between these telescoped autochthonous to parautochthonous settings and truly allochthonous accretionary complexes. Only carefully constructed structural-stratigraphic cross-sections, allowing some degree of palinspastic reconstruction, and underpinned by sufficient U-Pb zircon dating, can address the degree of allochthoneity of greenstone packages. Furthermore, seismic reflection profiles are essentially blind for the steep structures produced by multiple phases of upright folding and buoyant rise of mid- to lower-crustal, composite, granitoid and gneiss domes. Such structures are ubiquitous in granite-greenstone terrains and, indeed, most of these terrains appear to have experienced at least one phase of convective overturn to re-establish a stable density configuration, irrespective of the complexities of the pre-doming structural history. Buoyant rise of mid- to lower-crustal granitoid and gneiss domes can explain the typical size and spacing characteristics of such domes in granite-greenstone terranes, and the coeval deposition of late-kinematic, ‘Timiskaming-type’ conglomerate-sandstone successions in flanking basins. The extensional and subsequent contractional evolution of granite-greenstone terrains may have occurred in the overall context of a plate tectonic regime (e.g. volcanic rifted margins, back-arc basins) but highly extended, intraplate, rift-like settings seem equally plausible. Explaining the evolution of the latter in terms of Wilson cycles is misguided. Periods of intense rifting and flood volcanism (e.g. 2.73–2.70 Ga) may have been related to increased mantle plume activity or perhaps catastrophic mantle overturn events. Although there is evidence for plate-like lateral movement in late Archaean time (e.g. lateral heterogeneity of cratons, arc-like volcanism, cratonscale deformation patterns, strike-slip faults, etc.), the details of how these plate-like crustal blocks interacted and how they responded to rifting and collision appear to have differed significantly from those in Phanerozoic time. The most productive approach for Archaean research is probably to more fully understand and quantify these differences rather than the common emphasis on the superficial similarities with modern plate tectonics.