INTRODUCTION IMUs are accurate, mobile and cheap alternatives to motion capture cameras. The objective of this study was to evaluate the accuracy of foot-on/off detection in horses using different features derived from trunk and limb mounted IMUs. MATERIALS AND METHODS Four 18 g MTx IMUs were mounted on the fetlock of each limb. Five 10 g IMUs were mounted on the back. IMU data were sampled at 200Hz and synchronised to a 12 camera kinematic system and a force-plate runway. Data were processed with custom written Matlab scripts. The calibrated raw and rotated accelerometer, gyroscope and magnetometer, velocity and displacement data were analysed. RESULTS There was good front-limb (FL) foot-on accuracy and precision (mean less than 30ms bias and SD less than 50ms) for raw Z acceleration, Y angular velocity and X displacement. Z velocity had a poor accuracy (95ms bias) but high precision (SD 15ms). Hind limb (HL) foot-on accuracy and precision were good for raw X acceleration, rotated X acceleration, Z velocity, X displacement and Z velocity of the tuber coxae. FL foot-off accuracy was good for raw X and Z acceleration and rotated X acceleration, Y angular velocity and X velocity of the wither IMU, all with higher imprecision compared to foot-on (max. SD 60ms). For HL foot-off, accuracy and precision were good for Y angular velocity, Z velocity and X displacement and raw X acceleration of the tuber coxae. CONCLUSION IMUs provide an accurate, precise and repeatable method for detection of gait-events in horses.