Objective: To determine the efficacy of a modified constraint-induced therapy (mCIT) administered to patients with subacute stroke. Design: Prospective, multiple-baseline, before-after, randomized clinical trial. Setting: Subacute outpatient clinic. Subjects: Fourteen patients with subacute stroke who exhibited learned nonuse and stable motor deficits in their affected upper limbs. Intervention: Four patients participated in half-hour, structured physical and occupational therapy sessions that emphasized affected arm use in valued functional activities, 3 times per week for 10 weeks. Their less affected upper limbs were restrained 5 days per week during 5 hours identified as times of frequent use (mCIT). Five patients received regular therapy (TR) with similar therapeutic contact time to mCIT, and 5 patients received no therapy (CON). Main Outcome Measures: The Fugl-Meyer Assessment of Motor Recovery (Fugl), Action Research Arm (ARA) test, and Motor Activity Log (MAL). Results: After intervention, Fugl, ARA, and MAL scores remained virtually the same for TR and CON groups; scores improved by 11.4 and 11.5 points, respectively, on the Fugl and ARA for the mCIT group. Amount and quality of arm use, as measured by the MAL, also improved for mCIT patients (2.49 and 0.47, respectively). Conclusions: mCIT may be an efficacious method of improving affected arm function and use in stroke patients exhibiting learned nonuse.