The electromyographic (EMG) responses to tendon taps at the ankle and ensuing muscular twitch forces and temporal parameters were studied at varying angles across the joint range in 18 children, aged 3 to 14.9 years, with congenital hemiparetic cerebral palsy and 22 healthy (control) children, aged 3 to 13.6 years. Those subjects with hemiparesis were community ambulators without assistance. In all subjects, passive muscle stretch caused a waxing of the reflex EMG and twitch force near neutral (with the sole of the foot at right angles to the tibia) and a diminution of these with further dorsiflexion. Twitch times increased with each dorsiflexing increment, being slowest at maximum dorsiflexion and fastest at the resting plantarflexion angle. Heterogeneity of the hemiparetic-limb data is evident when compared with data of non-paretic and unaffected limbs, with clear differences in the clonic (fast twitch) as opposed to non-clonic (slow twitch) muscles. In four cases with clinical clonus, clonus frequency was reduced by passive dorsiflexion. Plaster immobilization for 1 month produced clonus which was previously absent in one subject, and caused a fast-twitch phenotype to emerge in two subjects. Follow-up after heel-cord lengthening in one subject showed that clonus frequency diminished from 9 to 3 Hz with slowing and strengthening of muscle-twitch phenotype. Short- and long-term peripheral manipulations appear to regulate neuromuscular excitability according to whether muscles are loaded or unloaded. Although damage to the nervous system provides the setting for reflex excitability, the data suggest that the muscle length (which specifies the joint angle) and the muscle-twitch phenotype of any given limb for any given case appear to dictate the actual speed and strength of reflex muscle-twitch and clonus profiles. This study illustrates how peripheral manipulations of muscles and tendons may alter the expression of what have hitherto been considered as exclusively central phenomena.