Robot-assisted gait training (RAGT) systems offer the advantages of standard rehabilitation and provide precise and quantifiable control of therapy. We examined the clinical outcome of RAGT and analyzed the correlations between gait analysis data and event-related desynchronization (ERD) and event-related synchronization (ERS) in patients with chronic stroke. We applied the Berg balance scale (BBS) and analyzed gait parameters and the ERD and ERS of self-paced voluntary leg movements performed by patients with chronic stroke before and after undergoing RAGT. A significant change was observed in BBS (p = 0.011). We also showed preliminary outcomes of changes in gait cycle duration (p = 0.015) and in ipsilesional ERS in the low-beta (p = 0.033) and high-beta (p = 0.034) frequency bands before and after RAGT. In addition, correlations were observed between BBS and ipsilesional ERS in the alpha and low-beta bands (r = −0.52, p = 0.039; r = −0.52, p = 0.040). The study demonstrated that RAGT can improve balance and provided an idea of the possible role of brain oscillation and clinical outcomes in affecting stroke rehabilitation.