Optically stimulated luminescence (OSL) dating studies of linear (longitudinal) dunes have been used extensively to elucidate late Quaternary environments and climates in arid or formerly arid regions, yet understanding of the development of such dunes is incomplete. In particular, conflicting opinions have been presented regarding the propensity of linear dunes to migrate laterally, the degree to which they rework their own sediment during accumulation and whether they form primarily by extension, as opposed to lateral sand movement from adjacent interdunes. This study focuses on this last point, although the importance of the other controversies is discussed in context. A simple linear dune in the south-western Kalahari, which has a prominent termination on a pan (playa) surface, provides an opportunity to directly test hypotheses of dune extension. Chronostratigraphy along a ~600 m transect along the crest of the dune, constrained by 42 OSL ages, reveals that the dune grew by extension on occasions in the late Pleistocene and early Holocene, but has also been subject to reworking along its length, which has continued until recent times. Dune development by extensional growth is suggested to operate under environmental conditions more conducive to net accumulation, whereas reworking is largely independent of conditions throughout the last ~18 ka, and may represent seasonal fluctuations in the position of the dune crest. The relative significance of these two modes of development is suggested to be a key control on the efficacy of linear dunes as archives of environmental and climatic change. Copyright © 2011 John Wiley & Sons, Ltd.