Background. Constraint-induced movement therapy (CIMT) has proven effective in increasing functional use of the affected arm in patients with chronic stroke. The mechanism of CIMT is not well understood. Objective. To demonstrate, in a proof-of-concept study, the feasibility of using kinematic measures in conjunction with clinical outcome measures to better understand the mechanism of recovery in chronic stroke patients with mild to moderate motor impairments who undergo CIMT. Methods. A total of 10 patients with chronic stroke were enrolled in a modified CIMT protocol over 2 weeks. Treatment response was assessed with the Action Research Arm Test (ARAT), the Upper-Extremity Fugl-Meyer score (FM-UE), and kinematic analysis of visually guided arm and wrist movements. All assessments were performed twice before the therapeutic intervention and once afterward. Results. There was a clinically meaningful improvement in ARAT from the second pre-CIMT session to the post-CIMT session compared with the change between the 2 pre-CIMT sessions. In contrast, FM-UE and kinematic measures showed no meaningful improvements. Conclusions. Functional improvement in the affected arm after CIMT in patients with chronic stroke appears to be mediated through compensatory strategies rather than a decrease in impairment or return to more normal motor control. We suggest that future large-scale studies of new interventions for neurorehabilitation track performance using kinematic analyses as well as clinical scales. © The Author(s) 2013.