Rehabilitation of locomotion in spinal cord (s.c.) injured patients is unsatisfactory. Here we report the effects of a novel 'Laufband (LB; treadmill) therapy' based on 'rules of spinal locomotion' derived from lower vertebrates. Eighty-nine incompletely paralysed (44 chronic and 45 acute) para- and tetraplegics underwent this therapy, then were compared with 64 patients (24 chronic and 40 acute) treated conventionally. The programme consisted of daily upright walking on a motor driven LB initially with body weight support (BWS) provided by a harness and assisted limb movements by the therapists when necessary. Forty-four chronic patients with different degrees of paralysis undertook the programme for 3-20 weeks (median = 10.5), 0.5-18 years after s.c. damage. At the onset of LB therapy 33/44 patients were wheelchair-bound (no standing and/or walking without help by others) whereas at the end of therapy 25 patients (76%) had learned to walk independently, 7 patients with help [corrected]. Only 1 subject did not improve. It was striking that voluntary muscle activity in the resting position was still low in several patients who had gained walking capability. Eleven patients who could already walk before LB therapy improved in speed and endurance. Of the 44 patients, six were capable of staircase walking before LB therapy compared with 34 afterwards. In order to validate the apparent superiority of LB therapy two types of comparisons were performed. In a 'temporal' control 12 spastic paretic patients, still wheelchair-bound after the period of postacute conventional therapy, performed LB immediately thereafter. After completion of LB therapy nine of these patients had learned to walk without help from others.(ABSTRACT TRUNCATED AT 250 WORDS)