Constraint-induced movement therapy (CIMT) has gained considerable popularity as a valuable treatment for a hemiparetic upper extremity. This approach is compatible with the emerging notion that task-oriented or functionally oriented retraining of the impaired limb provides evidence to support its utility. This article first provides a historical perspective on the development of CIMT. An overview model of how learned nonuse of the hemiparetic limb occurs and can be overcome with CIMT is discussed, and then a more detailed model that incorporates critical issues requiring considerably more basic and applied scientific exploration is described. Among the issues considered are the extent to which hemiparetic limb nonuse and subsequent modes of delivery to overcome it are governed by structure-function deficits rather than being attributable primarily to behavioral phenomena; the relative importance of the intensity of training; the need to better balance unimanual and bimanual upper-extremity task practice; the role of psychosocial and cultural factors in fostering patient compliance; the optimization of modes of delivery; and the reevaluation of the constellation of components contributing to successful outcomes with this treatment. Finally, the strengths, uncertainties, and limitations associated with CIMT are examined.