A newly developed quantitative method was used to measure spasticity in 18 patients with cerebral lesions and in 6 patients with spinal cord injuries. The flexor and extensor electromyographic (EMG) responses evoked by passive movement at the knee were compared. Contrary to classical theory, spasticity was more pronounced in the knee flexors than in the knee extensors of patients with hemiplegia as indicated by the stretch reflex threshold. Three types of flexor response patterns were evident in patients with hemiplegia but only one type was found in patients with paraplegia and quadriplegia. In addition to responses evoked during passive stretching, EMG activity was also evoked during passive muscle shortening. These ''shortening responses'' were found in most patients tested and were more evident in the flexors than the extensors. The clinical implications of these findings are discussed.