Objective: To examine the outcomes of home-based robot-guided therapy and compare it to laboratory-based robot-guided therapy for the treatment of impaired ankles in children with cerebral palsy. Design: A randomized comparative trial design comparing a home-based training group and a laboratory-based training group. Setting: Home versus laboratory within a research hospital. Participants: Children (N=41) with cerebral palsy who were at Gross Motor Function Classification System level I, II, or III were randomly assigned to 2 groups. Children in home-based and laboratory-based groups were 8.7 +/- 2.8 (n=23) and 10.7 +/- 6.0 (n=18) years old, respectively. Interventions: Six-week combined passive stretching and active movement intervention of impaired ankle in a laboratory or home environment using a portable rehabilitation robot. Main Outcome Measures: Active dorsiflexion range of motion (as the primary outcome), mobility (6-minute walk test and timed Up and Go test), balance (Pediatric Balance Scale), Selective Motor Control Assessment of the Lower Extremity, Modified Ashworth Scale (MAS) for spasticity, passive range of motion (PROM), strength, and joint stiffness. Results: Significant improvements were found for the home-based group in all biomechanical outcome measures except for PROM and all clinical outcome measures except the MAS. The laboratory-based group also showed significant improvements in all the biomechanical outcome measures and all clinical outcome measures except the MAS. There were no significant differences in the outcome measures between the 2 groups. Conclusions: These findings suggest that the translation of repetitive, goal-directed, biofeedback training through motivating games from the laboratory to the home environment is feasible. The benefits of home-based robot-guided therapy were similar to those of laboratory-based robot-guided therapy. (C) 2016 by the American Congress of Rehabilitation Medicine