Fencing, including electric fencing, is widely used across South Africa for livestock and game ranching practices. Leopard tortoises (Stigmochelys pardalis) are particularly prone to being killed by electric fences, but no published studies have assessed the impact of fence structure or quantified tortoise mortality along non-electric fences. This is a conservation concern, especially because South Africa is home to more tortoise species than any other country. This study relates tortoise mortality associated with electrified and non-electrified fences to fence structure, and uses transects away from fences as a comparison to estimate the impact of fences on mortality. All fence types had significantly higher tortoise mortality than open veld transects. Leopard tortoise mortalities were greatest along electric fences (56 % of mortalities), even though these comprised only 4% of fences in the study. By comparison, most angulate tortoises (Chersina angulata) died after becoming wedged in mesh fences. The distribution and abundance of fence types along 2200 km of roads was used to extrapolate the impacts of different fence types on tortoises in the southeastern Karoo, South Africa. A survey of land-use types indicated that game farms were more likely to be associated with the presence of electric fences. Regulations are needed to limit mortality of vulnerable species (tortoises, pangolins) on electric fences by setting a minimum strand height and ‘escape’ periods implemented through randomized off times or thermostatic switches. Keywords: Testudinidae, Chelonia, human-wildlife conflict, Stigmochelys pardalis, Chersina angulata, Psammobates tentorius.