Dynamic motor capacity was studied in 24 patients with spastic paraparesis (18 cases) or hemiparesis (6 cases). Torque was recorded with an isokinetic dynamometer in voluntary dynamic knee extensions and flexions at maximum effort and in passive movements at preset speeds of 30, 90 and 180 deg . s-1. EMG was recorded with surface electrodes from the quadriceps and the hamstring muscles. The capacity to accelerate motion up to preset speed was deficient in all patients as compared with healthy subjects. The moment of muscle force in movements at maximum effort was generally more reduced in fast than in slow movements. Spastic restraint in passive movements was low in the great majority of legs examined except in the fastest flexion, where about half showed restraint greater than 10 Nm. In a few cases, spastic reflexes were inhibited in voluntary motion. Most commonly, the restraint was greater in voluntary than in passive movements at equal range and speed. In voluntary motion, antagonist restraint was more common at high than at low speed of motion, and when present at low speed, it usually became greater with increasing speed of motion. It often reached a considerable magnitude, as estimated from EMG and may constitute a crucial component in the motor handicap.