In this laboratory we have developed a set of techniques that randomized controlled studies and a multisite randomized controlled trial have shown can substantially reduce the motor deficit of adult patients with mild to severe chronic strokes. Equivalent results have been obtained with adult patients after traumatic brain injury and brain resection. The basic technique, termed Constraint-Induced Movement therapy or CI therapy was derived directly from basic research with monkeys with mature motor systems and with monkeys given surgical intervention either on their day of birth or prenatally by intrauterine surgical procedures. We report here the results of two randomized controlled trials of CI therapy with young children with asymmetric upper extremity motor deficits of varied etiologies from 8 months to 8 years of age in one study and with children with hemiparesis consequent to prenatal, perinatal, or early antenatal stroke from 2 to 6 years old in a second study. The procedures used with children are very similar to those used with adults and diverge simply to make the basic techniques age-appropriate. All forms of CI therapy for the upper extremity to date involve 3 main elements: (1) intensive training of the more affected extremity, (2) prolonged restraint of the less affected extremity, (3) a 'transfer package' of techniques to induce transfer of therapeutic gains achieved in the laboratory to the life situation. The results in children with cerebral palsy are considerably better than those obtained in adults. Marked changes were observed in the quality of movement in the laboratory scored by masked observers from videotape; actual amount of use of the more affected arm in the life situation; active range of motion; and emergence of new classes of behaviour never performed before, such as in individual cases, fine thumb-forefinger grasp, supination, and use of the more affected extremity in crawling with palmar placement and rhythmic alteration. In the second experiment, the control group, after receiving usual and customary care for 6 months, was crossed over to receive CI therapy and exhibited results that were as good as those for the children receiving CI therapy first. Retention of treatment gains was approximately 70% at 6 months after the end of treatment. For some children there was no decrement in retention while for others there was a marked drop-off. One of the important factors contributing to good retention was the compliance of parents with the recommended post-treatment regimen. When retention is poor, brush-up periods may be of value. In the first experiment children were treated for 6 hr/day for 21 consecutive days, while in the second experiment treatment occurred only on the weekdays of the 3-wk treatment period (15 days). The results were at least as good with 15 days of treatment as with 21 consecutive days, thereby allowing the protocol to be fit into the usual therapist work week and making it more practical and less expensive for clinical use. CI therapy does not make movement normal in children with cerebral palsy with asymmetric upper extremity motor disorders. However, as carried out in this laboratory, it can produce a substantial improvement in a majority of cases.