BACKGROUND: Robotic-assisted total knee arthroplasty (RATKA) has been reported to enhance operative decision-making. The purpose of this study was intended to assess the predictive accuracy of bone cuts, lower limb alignment, and component size of a novel system for RATKA preoperatively and intraoperatively. METHODS: Preoperatively planned bone cuts, limb alignment, and component size were projected using a reconstructed 3D model. Intraoperative bone cuts and postoperative limb alignment were measured. Errors between planned and real bone cuts, limb alignment, and component size were compared. RESULTS: The mean absolute errors for bone cuts and alignment were within 1.40mm/1.30° with a standard deviation (SD) of 0.96mm/1.12°. For all errors of bone cuts and alignment compared with the plan, there were no statistically significant differences except for the lateral distal of femoral cuts (=0.004). The accuracy for predicting the femoral, tibial, and polyethylene component sizes was 100% (48/48), 90% (43/48), and 88% (42/48), respectively. Regarding all mean absolute errors of bone cuts and alignments, no significant differences were observed among surgeons. CONCLUSION: The novel robotically-assisted system for RATKA donated reliable operative decision-making based on the predictive accuracy regardless of the surgeon's level of experience.