Riding simulators are designed to replicate movement of a horse for the purpose of assessment and training of horse riders, but little is known about their similarity in replicating movement of horses. This study tested the validity of a dressage simulator, by measuring cycle/stride duration, range and symmetry of displacement of the simulator and comparing displacement vectors to that of real horses trotting on a treadmill. A reflective marker was placed on the midline of the simulator at the estimated level of the 18th thoracic vertebrae (T18), and over the T18 spinous process of ten horses. The simulator displacement was recorded in trot mode, while the real horses trotted at a comfortable speed on a treadmill. Displacements in three axes of motion were measured using 10 high-speed video cameras sampling at 240 Hz. Correlation tests showed high levels of statistical repeatability and symmetry of the simulator between multiple runs. Mean cycle/stride duration of the simulator was significantly faster than the group of horses by 0.17 s. Significant differences between the simulator and horses were shown in overall displacement in two axes, the simulator displaying 70% greater displacement in the mediolateral axis, 22% greater displacement in the craniocaudal axis, but displaying 12% less movement in the dorsoventral axis, which was not statistically significant. Displacement trajectories showed similarities in the frontal plane, displaying a butterfly-shaped sequence, but clear differences in the sagittal plane, with the horses showing an oval pattern of displacement and the simulator a clear linear displacement. Caution must therefore be taken with assumptions that riders will move in the same way on a simulator as they would on a real horse.