Sixty-six patients with end-stage knee osteoarthritis were randomized to receive EPMEDBOT robotic-assisted total knee arthroplasty (RATKA) (n=32) or manual jig-based TKA (n=34). The primary outcome was the proportion with mechanical axis malalignment not exceeding 3° (hip-knee-ankle, HKA). Secondary outcomes included Knee Society Score (KSS), WOMAC, range of motion (ROM), intraoperative/postoperative HKA deviation, component alignment angles (LTC, FFC, FTC, LFC), flexion–extension and medial–lateral gap metrics, operative time, blood loss, and complications. Outcomes were assessed within 3 days and at 3 months. Overall, the results showed RATKA achieved closer-to-target mechanical alignment and more accurate tibial/femoral component positioning, but longer operative/tourniquet times and no short-term differences in KSS, WOMAC, or ROM. In short, RATKA enhanced radiographic precision and intraoperative gap/alignment control without early functional superiority, suggesting potential long-term benefits that need confirmation with extended follow-up,