AbstractLand degradation in South Africa has been of concern for more than 100 years with both climate change and inappropriate land management (overgrazing) being proposed as primary drivers. However, there are few quantitative studies of degradation and, in particular, few of erosion by water. Badlands, taken here to be the landform which results from extreme erosion, have been notably neglected. We report on 13 consecutive years of erosion pin measurements of badland erosion on 10 study sites in the Sneeuberg uplands of the eastern Karoo in South Africa. The study sites are on Holocene colluvium which mantles footslopes. They have been subject to overgrazing for at least 100 years, c. 1850–1950. Currently they are lightly grazed by sheep. The area receives about 500 mm rainfall per year. The sites are remote, with only informal, farmer‐operated, daily raingauges nearby. The nearest sub‐daily raingauge is c. 55 km distant. Also we report on an analysis of the erosion pin data which focuses on establishing the origins and context of the badlands, including the relationship between study sites and adjacent valley‐bottom gully systems; compare erosion rates on our study sites with rates determined by erosion pins on other badland sites; and discuss the implications of these erosion rates for landscape development and off‐site impacts. Net erosion rates on the study sites are relatively high compared with global badland rates and range from 3.1 to 8.5 mm yr‐1 which may be extrapolated to 53 to 145 t ha yr‐1 (using a measured bulk density of 1.7 g cm‐3). However, comparisons with badland sites elsewhere are difficult because of different measuring methodologies, lithologies, climate and dominant processes. Erosion rates on the study sites are strongly influenced by rainfall amounts and, in particular, by daily rainfall events which exceed ~10 mm: this is the threshold intensity at which runoff has been observed to commence on badlands. Of significance, but of lesser influence, is weathering, mainly by wetting and drying: this prepares bare surfaces for erosion. However, questions remain regarding the role of site characteristics, and of processes at each site, in determining between‐site differences in erosion rate. Crude extrapolation of current rates of erosion, in conjunction with depths of incision into the badlands, suggests that badland development started around 200 years ago, probably as a response to the introduction of European‐style stock farming which resulted in overgrazing. We assume, but cannot quantify, the additional influence of periods of drought and burning in the erosional history of the area. Intermittent connection of these badlands to valley‐bottom gullies and therefore to small farm dams and ultimately to large water storage reservoirs increases their impact on local water resources. Copyright © 2015 John Wiley & Sons, Ltd.