Current rehabilitation methods fail to restore normal arm function for many stroke survivors, particularly those with severe deficits. The main objective of this study is to test efficacy and evaluate underlying neurophysiological mechanisms of a novel approach to treat persistent severe arm deficits after stroke with a combination of MyoPro™ and motor learning-based therapy. We will also estimate cost effectiveness of this therapeutic approach. Rationale: Motor learning-based therapy is one of the most effective stroke rehabilitation methods available, however its application is challenging for individuals with severe arm impairment because of their limited ability to practice volitional arm movement effectively. The MyoPro is an exoskeletal myoelectrically controlled orthotic device that is custom fitted to an individual’s paretic arm and assists the user to move the paretic arm. MyoPro can help with motor learning-based therapy for individuals with severe motor deficits as it motivates practice because even weak muscle activity is translated into patient-initiated arm movement. Preliminary results of motor-learning therapy using MyoPro in our laboratory showed an increase in Fugl-Meyer for Upper extremity score (FM) of 7.44 points following 18 weeks of training (18 in-clinic therapy sessions over 9 weeks followed by 9 weeks of home practice) for chronic stroke survivors with baseline FM≤30. However, comparison of the same dose of combination therapy with motor-learning alone remains to be determined. Study Design: Using a randomized, controlled design, individuals with chronic severe stroke (≥6 months post; Fugl Meyer UE score ≤30;n=60) will participate in either MyoPro+motor learning (M+ML) or motor learning alone (ML-alone). The study intervention will include 9 weeks of in-clinic training (18 sessions;1.5 hours each) followed by 9 weeks of home practice and a 6-week follow-up. Aim 1 is to determine whether M+ML results in greater treatment gains compared to ML-alone. The primary outcome will be change in FM. Secondary outcome measures will assess overall paretic arm performance and will include: kinematics, muscle tone (Modified Ashworth Scale; MAS), grip/pinch/arm dynamometry, sensory function (Semmes Weinstein mono-filament test, joint proprioception), arm function (Arm Motor Ability Test (AMAT);actigraphy) and quality of life (Stroke Impact Scale (SIS)). Aim 2 is to characterize structural and functional brain changes after treatment. Outcomes include corticospinal excitability (motor evoked potential recruitment curve (MEP-rc)), and functional connectivity (resting state function Magnetic Resonance Imaging(rs-fMRI). Aim 3 is to identify baseline factors associated with greater functional improvement with treatment. Outcomes are as follows: baseline integrity of the stroke-affected corticospinal tract (lesion load, MEP-rc; Diffusion Tensor Imaging); baseline motor ability of the affected arm (FM); baseline functional connectivity (rs-fM...