The influx in research to improve the evidence base and clinical effectiveness of existing practices has facilitated the emergence of new potential techniques. Constraint-induced therapy (CIT) is one such approach, used mostly widely in stroke rehabilitation. It focuses on the massed, random practice of functional tasks with the less able extremity, while the other is restrained. The evidence supporting the application of this technique is increasingly favourable, despite continued debate surrounding the optimal application. What remains scantly analysed are the underlying principles of CIT, to elucidate if massed, random practice of activities of daily living (ADLs) are the most suitable conditions to enhance learning. This paper helps to consolidate theoretically the advances made by existing clinical material. Although inhibiting initial performance, both massed and random practice provide greater learning potential. While further research is required, especially with these principles combined, this review concludes that CIT is based on the best available motor learning evidence, albeit the research has notable limitations.