An objective radiological method of assessing the degree to which horses are conformationally 'back at the knee' (hyper-extended) is described. The effects on the measurements of variations in the direction of the incident X-ray beam and variations in weight bearing by the horse were assessed. A change from a lateromedial projection towards a plamaro-lateral-dorsomedial oblique projection consistently tended to reduce the observed degree of hyperextension of the carpus. Raising the contralateral limb to increase the load on the carpus had little effect on the measurements. The carpi of 21 thoroughbred racehorses with carpal chip fractures were not significantly more hyperextended than those of 10 thoroughbred racehorses with normal carpi. Back at the knee conformation was unlikely to have played a major role in the aetiopathogenesis of the carpal injuries.