To explore the range of tensions used in reins to elicit specific movements from a range of horses, 22 horses of mixed age, sex, breed and training history were long-reined and ridden through a standard course. The reins contained embedded load cells so that tensions used to elicit specific movements could be measured and logged. These movements were categorised into 'left turn', 'right turn', 'going straight' and 'halt' and were separated for left and right rein tensions. The data were analysed using two-sample non-parametric Kolmogorov'Smirnoff tests and the differences between categories of horse and equipment were analysed with one-way analysis of variance. The tensions recorded in the reins were greater for long-reining than riding (median 5.76, Q25 3.9, Q75 13.3 N and median 5.29, Q25 9.3, Q75 2.9 N, respectively, P = 0.025), irrespective of whether the horses were ridden with a halter or a bridle or whether the test was completed at a walk or a trot. The tensions did not differ between the left and right reins (P > 0.05) when the horses were being driven or ridden in a straight line, providing evidence that an 'even contact' was maintained. The rein tension required for going straight was less than for any other responses, showing that a lighter contact on the reins can be maintained between the application of specific stimuli. The rein tension required to elicit the halt response was greater than for any other response (P < 0.001). The rein tensions required to complete the course did not differ with the use of bridle versus the halter (P > 0.05). Clearly, a range of rein tensions is required for horses to elicit specific responses. In the interests of horse welfare and avoidance of habituation, those involved in equitation need to become aware of the tensions used in training horses and seek to keep them to a minimum. When rein tension can be measured objectively, this process can be easily implemented and monitored.