Abstract: The behaviour of humans around horses is thought to have a substantial impact on how people are perceived in subsequent interactions and many horse trainers give detailed advice on how handlers should behave when initially approaching a loose horse. Here we report on three studies designed to explore the effect of different human approach styles on the behaviour of naïve and experienced horses. In the first study, the change in flight distance (distance at which horses started to avoid an approaching human) of twelve semi-feral Dartmoor ponies, undergoing training to allow handling, was assessed. Over the 10 handling sessions median flight distance decreased significantly (p < 0.001) from 2.38 m to 0.00 m and there was a significant positive shift in the ponies' behaviour following the appearance of the researcher (p = 0.002). In a second study the effect of a direct (vigorous, swinging a lead rope and with eye contact) versus indirect (relaxed, no rope swinging and without eye contact) approach style was assessed on six adult experienced riding horses. The mean flight distance during a direct approach style (6.87 m) was significantly greater than that which occurred during an indirect approach style (2.32 m). Direction of approach was not found to significantly affect flight distance. In a third study, the effect of the rope was removed and a similar method to the second study applied to a group of naïve, feral ponies. The effect of different components of approach style, speed of approach, handler body posture and direction of gaze, which might contribute to observed differences in behavioural responses, were then examined systematically in this population. This revealed no significant difference in mean flight distance between the two approach styles (2.28 m indirect versus 2.37 m direct approach), but ponies were significantly more likely to move off in trot (p = 0.025) and to travel further (p = 0.001) when a direct approach was used. Speed of approach was the most salient factor, with a fast approach increasing both the tendency to move off in trot (p < 0.001) and distance travelled (p < 0.001). Body posture (relaxed or tense) had no effect, while flight distance was significantly greater when the person was looking away (p = 0.045). These results suggest horses may have an important egocentric spatial barrier, which perhaps relates to personal space and triggering of the flight response. Contrary to popular belief, body posture did not appear to be very important in the contexts examined unless accompanied by extraneous aids, while the speed of approach is particularly significant. These results are of important practical relevance in reducing the risk of injury, and the effective management of horses with minimal stress.