In this study the influence of voluntary upper body exercise on the performance of stimulated paralysed human quadriceps was investigated in five subjects with spinal cord lesions in the thoracic spine. The experimental setup consisted of computer-controlled stimulation of the quadriceps using electrodes on the surface of the skin, a dynamometer for isometric or isokinetic loading of the lower leg, and a rowing ergometer for upper body exercise. In all subjects, quadriceps fatigue tests were conducted to study the influence of upper body exercise on knee torque during sustained continuous or intermittent stimulation of quadriceps. The relative asymptotic torque appeared to be significantly higher with the presence of upper body exercise than without. This was consistently found both between trials (starting with or without upper body exercise) as well as within trials, when upper body exercise was started or stopped during the trial. No significant influence of upper body exercise on the time constant of initial torque decline was found.