The authors reported the changes of spinal cord evoked potential following impact injury by the weight dropping technique in untreated and treated animals. The correlation between the change in amplitude of the spinal cord evoked potential after injury and the prognosis for motor recovery of legs was also discussed. (Materials and Methods) The spinal cord of 57 adult dogs was traumatized by the weight dropping technique at the level of Th10. 37 dogs were not treated. The trauma consisted of 300 gm-cm impact injury, 400 gm-cm impact injury and 500 gm-cm impact injury. The motor function of the dogs was observed by clinical scale for rating hind limb motor function (modified Tarlov score) for one month after trauma. 20 dogs with 300 gm-cm contusion were treated with mannitol. 10 dogs were treated with intravenous continuous infusion of mannitol (1 g/kg/hr) 30 minutes after injury and 10 dogs with intravenous bolus injection of mannitol (2 g/kg) at 30 minutes and 3 hours after injury. The spinal cord evoked potential following direct stimulation of the spinal cord was recorded from the epidural space for 5 hours. The recording electrode was located cephalad to the site of the impact injury. (Results and Conclusion) 1. In each group, the amplitude of I potential decreased remarkably immediately after injury. Subsequently, gradual recovery was obtained. The spinal cord evoked potential following impact injury by the weight dropping technique could be estimated one hour after trauma in 300 gm-cm, 30 minutes after trauma in 400 gm-cm injury and immediately after trauma in 500 gm-cm injury. 2. Spinal cord evoked potential was useful in monitoring the recovery potential of the spinal cord. Recovery of amplitude more than 40% indicated a favorable prognosis. On the other hand, recovery of amplitude less than 19% showed a poor prognosis. However, the precise prognosis for motor recovery could not be accurately estimated in dogs with recovery of amplitude between 39% and 20%. 3. In mannitol-treated group with intravenous continuous infusion, the recovery of the amplitude was superior to that in untreated group at a significance level of p less than 0.05. However, in the group with intravenous bolus injection of mannitol, the recovery was transient.