Open-air archaeology plays a limited role in southern African Late Pleistocene research, with most studies focused on rock shelter assemblages. Recently, archaeologists have noted discrepancies in the composition of Late Pleistocene lithic assemblages between some of the region’s open-air and rock shelter sites. For example, although relatively abundant in rock shelters, Late Pleistocene Later Stone Age (LSA, c. 44–12 kcal. BP) bipolar cores are rare in open-air contexts. In this paper, we assess this discrepancy by testing for differential preservation of specific artefact classes and sizes in semi-arid open-air conditions. We placed a replicated assemblage of miniaturised cores and flakes on an archaeologically sterile sediment surface in the Doring River Valley (South Africa) and recorded their movements over 22 months. Our results indicate that bipolar and freehand cores moved comparable distances within the study interval and that surface slope is the strongest predictor of miniaturised tool movement. We also show that (1) relatively flat lithics move disproportionately more and (2) random artefact orientations do not preclude local (i.e. metre) scale artefact transport. In terms of the archaeology of our study area, the observed clustering of surface artefacts on sediment bodies likely results from their recent exposure. Our data suggest that the paucity of open-air bipolar artefacts in Late Pleistocene LSA assemblages may have more to do with human behavioural variability at landscape scales than differential preservation. Southern Africa’s rich rock shelter record is, therefore, unlikely to represent the full suite of prehistoric hunter-gatherer behaviours.