Question: What is the effect of constraint-induced movement therapy (CIMT) on upper limb function in patients 3 to 9 months after stroke? Design: Randomised controlled trial with concealed allocation and assessor blinding. Setting: Seven universities in the USA, recruiting participants from 247 medical facilities. Participants: 222 adults 3 to 9 months after their first ischaemic or haemorrhagic stroke with at least 10° of active extension at the wrist and at the thumb and two fingers. Participants also had to demonstrate adequate balance, transfer, sit-to-stand, and standing ability while wearing the mitt used to apply CIMT. One hundred and six patients were randomised to CIMT and 116 to usual care. Interventions: CIMT involved wearing a restraining mitt on the less-affected hand for 90% of their waking hours over a 2-week period. On weekdays during this period, repetitive task practice and behavioural shaping with the hemiplegic hand were performed for up to 6 hours per day, with 30 minutes of additional practice of the tasks at home. Usual care ranged from no treatment after concluding formal rehabilitation to the application of orthotics or occupational/ physical therapy on a domiciliary or outpatient basis. Outcomes: The primary outcomes were the Wolf Motor Function Test (WMFT), which includes measures of speed and strength of upper extremity motor function, and the Motor Activity Log (MAL), an interview-derived measure on a 0-5 scale of the amount and quality of performance of 30 common daily activities. The Stroke Impact Scale (SIS) health status interview was a secondary outcome measure. All outcomes were measured after treatment and at 4, 8, and 12 months post-treatment. Results: From baseline to 12 months, the CIMT group showed greater improvements than the usual care group in: the WMFT Performance Time by 34% (95% confidence interval (CI) 12 to 51); the MAL Amount of Use by 0.43 (95% CI 0.05 to 0.80); and the MAL Quality of Movement by 0.48 (95% CI 0.13 to 0.84). The CIMT group also had a significantly greater decrease in self-perceived hand function difficulty on the SIS hand domain, by 9% (95% CI 0.3 to 19). Conclusion: Among patients who have had a stroke in the past 3 to 9 months, CIMT produces improvements in arm motor function that persist for at least one year. [ABSTRACT FROM AUTHOR]