Purpose To determine whether gait training with a hybrid assistive limb (HAL) as an exoskeleton robotic device was safe and could increase functional mobility and gait ability in subacute stroke patients. Methods The participants were eight patients with post-stroke hemiparesis whose walking impairment and gait recovery curves had plateaued. The intervention program was gait training using HAL and a walker for 20 min daily 5 days a week for 5 weeks. The 10-m maximum walking speed (MWS), self-selected walking speed (SWS) and 2-min-walk test (2MT) without HAL were used as primary outcome measures to determine the effects of training. The Berg Balance Scale (BBS), Fugl-Meyer Assessment of Motor Recovery after Stroke (FMA) and functional ambulation category (FAC) were assessed as secondary outcomes. These measures were assessed before and after the 5-week intervention program and were analyzed statistically using a paired t-test. Results All eight participants completed the intervention program with no adverse events. There were significant increases in MWS, SWS and 2MT. BBS, FMA and FAC also increased, but not significantly. Conclusion The new HAL exoskeleton robotic device was efficient and safe for improving motor function and gait in patients in the subacute stage after stroke. Implications for Rehabilitation Gait training using HAL will improve gait ability of individuals with post stroke. The HAL achieve intensive gait training without increase spasticity and abnormal gait pattern.