Background/Purpose: Physical rehabilitation of walking post-stroke is designed to improve individuals' mobility in the presence of motor, sensory and balance impairments. Effective strategies to improve balance and decrease fall risk continue to be a need in stroke rehabilitation. To address this, our laboratory has studied a novel Backward Walking Training program (BWTraining) to improve both forward and backward walking speed as well as balance in individuals with both acute and chronic stroke. Our prior published work has demonstrated that this intervention is safe, feasible and effective, however, critical questions remain unanswered regarding how best to deliver this intervention for optimal impact. The primary aim of this proposal is to establish dosage effects of BWTraining for maximizing gains in forward walking speed, backward walking speed, and dynamic balance. Additionally, we will determine baseline brain MRI predictors of response to BWTraining intervention and explore changes in brain activity as a result of this behavioral intervention. We are interested in understanding how individual differences in structural and functional brain measurements might distinguish between responses to BWTraining. This information will promote prudent decision-making as to which patients are good candidates for this novel intervention. We will conduct a randomized controlled trial with three specific aims to test the 1) Dose effect of number of BWTraining sessions on walking speed and balance, 2) the ability of resting state Functional Connectivity (rs-FC) between specified brain regions and Blood Oxygen Level-Dependent (BOLD) activation of brain Regions Of Interest (ROIs) during task-based fMRI to predict BWTraining outcomes and 3) changes in both rs-FC and BOLD activation during contralateral ankle motion following BWTraining intervention compared to baseline. Study Design: This is a randomized single-blind controlled trial. Subjects: Forty-eight individuals, 2-4 months post-stroke, will meet the following study criteria: 1) ambulatory with gait speed < 0.8 m/s; 2) community-dwelling; 3) Berg Balance Scale < 42; 4) absence of other neurological conditions; 5) stable cardiac status; 6) absence of lower extremity orthopedic impairments or pain that limits gait ability; 7) absence of metal and Magnetic Resonance sensitive implanted medical devices. Methods: Following baseline Assessment A, participants will be randomized to an 18 or 27 session BWTraining group. BWTraining intervention (3X/week for 6 or 9 weeks) consists of 20–30 minutes of step training using a Body Weight Support and Treadmill system followed by 20 minutes of overground gait training. Assessment B will occur after the final training session for each group. Assessment C will occur 6-weeks post- intervention and Assessment D at six months post-intervention for both groups. Magnetic Resonance Imaging of the brain consisting of rs-FC and BOLD blood flow in ROI's during an ankle motor task...