In 44 patients having cordotomies for relief of the pain of cancer, a correlation was made between the location and extent of the incision in the spinal cord and the motor state. Post-mortem histological examination of the spinal cord was carried out in all areas. An incision cutting through one anterior quadrant of the cord at any segmental level could be made without causing any disturbance of motility. An incision in the thoracic segments cutting through the anterior half of the cord could be made without causing any disturbance of motility. This fact implies that tracts in the posterior half of the cord can supply the input from the brain necessary for the maintenance of functions normally mediated by tracts in the anterior half of the cord. The more posterior the incision reached in the posterolateral column, the greater were the defects in motility. A large unilateral lesion dividing most of the lateral corticospinal tract, and the descending fibres anterior to it, caused flaccid paralysis of the ipsilateral lower limb. Voluntary movements started to return within 5 h. An incision in the thoracic cord cutting through one lateral corticospinal tract and 85-90% of the opposite tract and reticulospinal fibres anterior to that tract caused total paralysis of the lower limbs. Recovery ensued over 2 months so that the patient eventually walked, though with severe spastic paraparesis. Recovery of some flexor and extensor movements of the ipsilateral fingers and toes occurred within 6 h of an incision being made in the upper cervical cord that divided the lateral corticospinal tract unilaterally. Division of only the anterior fibres of the lateral corticospinal tract above the cervical enlargement did not affect the motility of the ipsilateral upper limb. It is concluded that in the more cranial segments of the spinal cord, corticospinal fibres destined for the upper limb are in the more posterior part of the tract. Correlation of the clinical with the histological evidence of a lesion of the lateral corticospinal tract was carried out. When it was deduced on the clinical evidence that the tract was damaged, this was always found to be correct. On the other hand, the tract might show histological evidence of damage without manifesting any evidence of a lesion. The Babinski response was found, in general, to occur with lesions of the lateral corticospinal tract and not with lesions elsewhere in the cord.(ABSTRACT TRUNCATED AT 400 WORDS)