The aim of this investigation was to study the effect of non-involved-side exercise on the involved (paretic) leg-muscle of patients with stroke. Each of the ten adults studied attempted to perform, while seated, a predetermined sequence of the following actions : (1) a voluntary knee extension on the involved side; (2) a maximal, isometric knee-flexion on the non-involved side; (3) a maximal, isometric knee-flexion on the non-involved side with counter-resistance (in front of the ankle joint) on the involved side. The root-mean-square voltage was measured across the rectus femoris, the medial hamstring, the tibialis anterior, and the medial gastrocnemius muscles (on the involved side) with surface electrodes. The root-mean-square voltages of the involved-side rectus femoris and the tibialis anterior muscles were found to increase substantially during non-involved-side knee flexion relative to that detected for voluntary knee extension on the involved side. The voltage of the rectus femoris muscle increased substantially also when counter-resistance was present. There was no similar substantial increase of muscular activity, in the medial hamstring and medial gastrocnemius, induced by contralateral isometric knee-flexion. This study suggests that, among patients with severe hemiplegia, isometric knee flexion in the non-involved side may be useful for facilitating the paretic rectus femoris and tibialis anterior muscular activities.