Fences are utilised throughout the world to restrict the movements of wildlife, protecting them from threats and reducing human—wildlife conflict. In South Africa the number of privately-owned fenced game reserves has greatly increased in recent years, but little is known about how fencing affects the distribution and movements of target and non-target mammals. We surveyed 2 m either side of the complete fence line of a recently established commercial game reserve in South Africa, identifying signs of animal presence (spoor, scat, foraging or other field signs) while also recording damage (holes) to the fence. Every 250 m we carried out 100 m perpendicular transects either side of the fence, recording vegetation cover and height at 10 m intervals along the transect. We found that livestock (largely cattle) were excluded from the reserve. However, 12% of records of large animal species were recorded outside of the fence line. These species had been introduced to the reserve, strongly suggesting that they had crossed the boundary into the surrounding farmland. Sixteen naturally present wild species were found on both sides of the fence, but we found more evidence of their presence inside the reserve. Observational evidence suggests that they were regularly crossing the boundary, particularly where the fence was damaged, with hole size affecting species recorded. We also found evidence that the construction of the fence had led to a difference in vegetation structure with plant richness and percentage of non-woody plant cover significantly higher inside the fence. While fencing was highly effective at preventing movement of livestock, introduced and wild animals were able to cross the boundary, via holes in the fence. This work shows that the efficacy of the most common approach to preventing animal movement around protected areas depends on the species being considered and fence condition.