I examined the isokinetic knee extension mean torque in 30 patients with hemiplegia and 30 healthy subjects matched by age, sex, height, and weight at knee extension velocities of $30^{circ},;60^{circ},;90^{circ}$, and $120^{circ}$/sec. The purpose of this study is as follows: first, to compare isokinetic data between the involved and uninvolved side of patients and healthy subjects. secondly, to determine whether the relative decreases in knee extension mean torque at velocities greater than $30^{circ}$/sec were different on the two sides. Mean torque of quadriceps on both sides of patients with hemiplegia was significantly less than the mean torque of muscle of healthy subjects at various speeds. Relative decreases in knee extension mean torque differed between speeds, but not between sides. Torque at speed greater than $30^{circ}$/sec were correlated significantly with the torque at $30^{circ}$0/sec. These finding indicated that isokinetic testing can provide objective information about quadriceps muscle performance in hemiplegia, and suggested that hemiplegia may have difficulty in moving forcefully at higher speeds than $30^{circ}$/sec because they are weak. Therapeutic intervention, therefore, might be most beneficial when they are directed toward helping patients with hemiplegia activate their muscle weakness.