The Transvaal Basin in South Africa hosts a 15 km thick pile of sedimentary successions deposited over a period of >600 Ma during the Neoarchean to Paleoproterozoic. Presently, little is known about the source of these sediments, as well as about the tectono-magmatic evolution in the hinterland of the Transvaal Basin, preventing detailed geotectonic correlations of the Kaapvaal Craton (KC) with other cratons worldwide. To solve this problem, we present the first systematic study of combined U-Pb and Lu-Hf isotope data of >2000 detrital zircons from fourteen formations of the Transvaal Supergroup. These reveal that clastic sedimentary rocks were supplied from sources on and off the present-day KC. Detrital zircons in conglomerates of the Wolkberg and Black Reef formations, maximum deposition ages at 2769 ± 8 and 2618 ± 11 Ma respectively, were mainly supplied from surrounding KC, either from Pietersburg Block Basement (PBB), and/or from eroded sedimentary successions of the Witwatersrand, Pongola and/or Ventersdorp Supergroups. In contrast, clastic sedimentary rocks of the Rooihoogte, Duitschland and Timeball Hill formations (maximum deposition ages at 2353 ± 18 Ma, 2342 ± 18 and 2290 ± 8 Ma, respectively) were predominately supplied from a juvenile Neoarchean terrane (JUNAT) formed at 2570–2500 Ma (εHf2500 Ma = +2 to +9) and intensely reworked at 2400 Ma, and to a minor amount from a composite Archean terrane (CAT) emplaced by granitoids between 3540 and 2680 Ma, and affected by crust reworking at 2570–2430 Ma (εHf2.5Ga = −3 to −12) in a Neoarchean to Paleoproterozoic continental arc terrane (NPCAT). Subsequent periodic reworking of JUNAT at 2250–2220 Ma and 2120 Ma is recorded by detrital zircons in sandstones of the overlying Boshoek, Dwaalheuwel, Daspoort, Magaliesberg and post-Magaliesberg formations, having maximum deposition ages at 2243 ± 7, 2242 ± 7, 2240 ± 7, 2080 ± 7, and 2068 ± 7 Ma, respectively. The Archean zircons (age >2650 Ma) in all these formations were mainly supplied from PBB. The new data sets also suggest that the KC was connected to CAT, NPCAT and JUNAT at <2350 Ma. The nearly absence of detrital zircons with ages of 2570–2500 Ma in all formations younger than Boshoek perhaps results from intense reworking of JUNAT during magmatic events at 2400, 2340, 2220, and 2120 Ma, causing loss of the original juvenile character. Paleoproterozoic zircons with ages of 2220 and 2120 Ma in Dullstroom sandstones most likely result from re-deposition of post-Magalisberg sedimentary rocks, and Archean zircons from sources similar to Moodies and Fig Tree sandstones of the Barberton greenstone belt. Comparison of our new data from the Transvaal Basin with such from the Turee Creek and Horseshoe basins in NW-Australia provides no evidence for Kaapvaal-Pilbara Craton connection during the Neoarchean to Paleoproterozoic.