: The ROSA Knee System (Zimmer Biomet, Warsaw, IN, USA) is a robotic system aiming to increase bone resections and component alignment accuracy during TKA. While much is known about its performance in the coronal plane, its accuracy in the sagittal plane remains debated. The present investigation evaluated the system's accuracy in achieving planned mechanical axis alignment and specific knee angles in both planes. A retrospective analysis was performed on 55 consecutive patients who underwent robotic-assisted TKA using the ROSA Knee System. Data on the medial proximal tibial angle (MPTA), lateral distal femoral angle (LDFA), hip-knee-ankle angle (HKA), tibial slope (TS), and distal femoral flexion (DFF) were collected pre- and post-operatively using the ROSA software. Planned and achieved angles were compared, with deviations greater than 2° and 3° defined as outliers. The mean differences between planned and achieved angles for LDFA and MPTA were 0.5° ± 1.00° and 0.3° ± 1.3°, respectively, with less than 10% outliers. The hip-knee angle recorded only a minimal deviation from planned values. In contrast, the TS angle showed a statistically significant difference between planned and achieved values, while no significant difference was found for the DFF angle. The surgeon's experience did not impact alignment accuracy. The ROSA Knee System demonstrates high accuracy in achieving planned alignment in the coronal plane during robotic-assisted TKA, with minimal outliers and reliable predictions for both femoral and tibial angles. However, the ROSA Knee System showed less accuracy in the sagittal plane, particularly for the tibial slope, which did not adversely affect the implant's stability.