The late Archaean--early Proterozoic Witwatersrand basin is situated in the center of the Kaapvaal craton, South Africa. Two new heat flow determinations and 71 new heat flow estimates permit excellent definition of the heat flow pattern. Nonvertical heat conduction explains much of the observed local heat flow variability. Regionally, the heat flow in the basin is uniform, and the overall mean (51¿6 mW m-2) is higher than the mean heat flow for Archaean cratons. Published data indicate that heat generated in the basin's strata (~7.5 mW m-2) is largely responsible for the excess heat flow; the subsediment heat flux is ~43.5 mW m-2. Seven new heat flow and heat production determinations in Archaean granite domes northwest of the basin average 33¿2 mW m-2 and 2.3¿0.8 μW m-3, respectively. Contrasted with the heat flux across the floor of the basin, the lower heat flow in the domes probably reflects a smaller radiogenic contribution from the crystalline crust below the domes. Published heat production data from upturned granitic basement exposed in the Vredefort dome in the center of the basin indicate that the radioactive contribution from the crust below the basin is ~26.5 mW m-2. Subtracting this value and the contribution from the strata (~7.5 mW m-2) from the basin's surface heat flow (51 mW m-2) indicates that the heat flux through the Moho is approximately 17 mW m-2. Calculated lithospheric temperatures agree closely with pressure-temperature regimes inferred from studies of mantle inclusions in kimberlites in the Kaapvaal craton. Both data sets indicate that the cratonic lithosphere is considerably cooler and thicker than that below an equilibrium ocean. Implications of the cold, thick cratonic lithosphere for understanding the distribution of southern African diamondiferous kimberlites, and the density and buoyancy of the lithosphere, are discussed. ¿ American Geophysical Union 1988