Twenty spastic cerebral palsy (CP) patients undergoing chronic cerebellar stimulation (CCS) for reduction of spasticity and improvement in function have participated in a double-blind study. Seven US centers involving 9 neurosurgeons (1984-6) have replaced the depleted Neurolith 601 fully implantable pulse generator (Pacesetter Systems Incorp.-Neurodyne Corp., Sylmar, CA) with new units in 19 CP patients, 1 patient entered the study following his initial implant. A magnetically controllable switch was placed in line between the Neurolith stimulator and the cerebellar lead, so allowing switching sequences for the study. Physical therapists, living in the vicinity of the patient's home, carried out two quantitative evaluations: 1. Joint angle motion measurements (passive and active). 2. Motor performance testing was done when possible and included: reaction time, hand dynamonetry, grooved peg board placement, hand/foot tapping, and rotary pursuit testing. Testing was done presurgery, at 2 weeks postimplant, then the switch was activated either "on" or "off" to a schedule, with testing and reswitching at 1, 2 and 4 months, then the switch was left turned "on". Of the 20 patients, 16 finished the tests, 2 patients failed to finish and 2 had switch problems and were deleted from the study. Two of the 16 patients were "off" through the entire testing. Of the 14 that had periods of the stimulator being "on", 10 patients (72%) had quantitative improvements of over 20%, (1 pt: 50+% improvements; 4 pts: 30-50%, 5 pts: 20-30%); while 1 patient (7%) had improvements in the 10-20% level, whereas 3 patients (21%) showed no improvement.