Many of the models for modern sedimentary basins postulate two-stage subsidence; a rapid initial subsidence due to thinning or loading of the crust, followed by a more protracted thermal stage as the lithosphere, which is thinned during the initial stage, relaxes to equilibrium thickness. The geology of a number of Archaean green-stone belts and early Proterozoic cratonic basins in South Africa may be explained by such a model. Rapidly erupted shallow marine or subaerial volcanic rocks predominate in the lower parts of sedimentary--volcanic sequences. These are thought to relate to initial subsidence as (1) accommodation of relatively thick volcanic sequences requires substantial and rapid subsidence, and (2) marginal uplift following these early volcanic intervals is consistent with viscous relaxation following the initial elastic response of the lithosphere to localized loading. Sedimentary sequences overlying initial volcanic dominated intervals may have been deposited during the ensuing phase of more widespread subsidence related to thermal relaxation of the thinned lithosphere. If so, sediment-filled subsidence of ca. 5.5 km in greenstone terrains at 3.5 and 2.6 Ga and of 7-10 km in cratonic shelf basins between 2.7 and 2.1 Ga require increases of lithosphere thickness between ca. 60 and 90 km. These minimum estimates of early Precambrian lithosphere thickness, although crude, are similar to estimates of present lithosphere thickness. In some early Precambrian basins, the cause of subsidence may have been crustal extension with development of faulted grabens that evolved into continental margins, but in cratonic shelf basins faulting did not occur during or after the initial subsidence, and some less obvious causal mechanism must be sought.