One of the unique spots in our Earth is the Kaapvaal craton; it preserves a rich record of the early earth and is underlain by unusually thick, ~ 200 km deep, chemically depleted lithospheric mantle of early Archean age. The internal structure of this unusual upper mantle, termed the tectosphere to 350 km depth by Jordan (1975), has to be revealed in more detail, and a better understanding should yield new insight into the origin of Earth’s early continents. We have reassessed the data from the Kaapvaal seismic experiment for lithosphere structure by application of PdS receiver functions analysis. Our main emphasis has been on the Kaapvaal craton and its boundaries with the surrounding tectonic belts. The analysis includes data from 95 temporary and permanent stations. The epicentral distances are between 30 and 95 degrees and the magnitudes range between 5.5 to 8.1 Mw, corresponding to a total of 534 events during the two years of recording from 1997 to 1999. We give special emphasis to imaging converters in the upper mantle, where crustal multiples often coincide with the conversions from real mantle boundaries. By modelling predicted multiples for comparison with the PdS receiver function image, we find that the receiver functions indicate the presence of two low-velocity zones (LVZ) between 170 and 225 km depth and below 350 km depth. We are currently processing SdP receiver functions to provide another constraint on these features. Modelling further indicates the presence of a significant difference in anisotropy between the craton and the surrounding tectonic belts. The evidence for this variation arises from the good azimuthal coverage, which provides an obvious polarity reversal of signals versus back azimuth coinciding spatially with the Kaapvaal boundaries. To confirm these preliminary results, we are continuing the experiments by calculation of theoretical receiver functions for a range of models, and are assessing the combined integrated PdS and SdP receiver function response in combination with teleseismic tomography to provide an integrated high resolution model.