Objective: Constraint-induced movement therapy (CIMT) is a method to improve motor function in the upper extremity following stroke. The aim of this trial was to determine the effect and feasibility of CIMT compared with traditional rehabilitation in short and long term. Design: A randomized controlled trial. Setting: An inpatient rehabilitation clinic. Subjects: Thirty patients with unilateral hand impairment after stroke. Intervention: Six hours arm therapy for 10 consecutive weekdays, while using a restraining mitten on the unaffected hand. Main measures: The patients were assessed at baseline, post-treatment and at six-month follow-up using the Wolf Motor Function Test as primary outcome measure and the Motor Activity Log, Functional Independence Measure and Stroke Impact Scale as secondary measurements. Results: The CIMT group (n=18) showed a statistically significant shorter performance time (4.76 seconds versus 7.61 seconds, P= 0.030) and greater functional ability (3.85 versus 3.47, P= 0.037) than the control group (n=12) on the Wolf Motor Function Test at post-treatment assessment. There was a non-significant trend toward greater amount of use (2.47 versus 1.97, P= 0.097) and better quality of movement (2.45 versus 2.12, P=0.105) in the CIMT group according to the Motor Activity Log. No such differences were seen on Functional Independence Measure at the same time. At six-month follow-up the CIMT group maintained their improvement, but as the control group improved even more, there were no significant differences between the groups on any measurements. Conclusions: CIMT seems to be an effective and feasible method to improve motor function in the short term, but no long-term effect was found.