Abstract Electromechanical-assisted gait training may be an effective intervention to promote motor recovery after brain injury. However, many studies still have difficulties in clarifying the difference between electromechanical-assisted gait training and conventional gait training. To evaluate the effectiveness of electromechanical-assisted gait training compared to that of conventional gait training on clinical walking function and gait symmetry of stroke patients. We randomly assigned patients with stroke ( n = 144) to a control group (physical therapist-assisted gait training) and an experimental group (electromechanical gait training). Both types of gait training were done for 30 min each day, 5 days a week for 4 weeks. The primary endpoint was the change in functional ambulatory category (FAC). Secondary endpoints were clinical walking functions and gait symmetries of swing time and step length. All outcomes were measured at baseline (pre-intervention) and at 4 weeks after the baseline (post-intervention). FAC showed significant improvement after the intervention, as did clinical walking functions, in both groups. The step-length asymmetry improved in the control group, but that in the experimental group and the swing-time asymmetry in both groups did not show significant improvement. In the subgroup analysis of stroke duration of 90 days, FAC and clinical walking functions showed more significant improvement in the subacute group than in the chronic group. However, gait symmetries did not show any significant changes in either the subacute or the chronic group. Electromechanically assisted gait training by EXOWALK was as effective as conventional gait training with a physiotherapist. Although clinical walking function in the subacute group improved more than in the chronic group, gait asymmetry did not improve for either group after gait training. Trial registration : KCT0003411 Clinical Research Information Service (CRIS), Republic of Korea.