Purpose: Implant malpositioning, joint line (JL) lowering, and malalignment have been identified as risk factors for implant failure in unicompartmental knee arthroplasty (UKA). The aims of this study were to compare the accuracy of implant positioning in robotic-assisted UKA versus conventional UKA in a large cohort and examine the correlation with implant survival at mid-term follow-up. Methods: This retrospective study included 356 medial UKAs from 2011 to 2019. The radiological measurements performed were coronal positioning of tibial implant according to Cartier angle (Δ Cartier), posterior tibial slope (PTS), residual hipknee-ankle (HKA), and JL restoration. Outliers were defined as follows: post-operative HKA < 175° or > 180°, Δ Cartier > 3° or < − 3°, JL change ≥ 2 mm, and PTS < 2° or > 8°. The survival probability was reported at the last follow-up. Results Out of the 356 knees included, 159 underwent conventional UKA (44.5%) and 197 (55.5%) robotic-assisted UKA. The mean follow-up was 61.3 months ± 24.0. Robotic UKA was associated with better accuracy compared to conventional UKA in relation to HKA (67% vs 56%, p = 0.023), JL restoration (70% vs 44%, p < 0.0001), PTS (83% vs 55%, p < 0.0001), and tibial varus restoration (65% vs 55%, p = 0.049). Implant survival in the robotic group was found to be superior at the last follow-up (96.4% versus 87.3% at 9 years, p = 0.004). Conclusion Robotic assistance in patients undergoing medial UKA was associated with better accuracy compared to conventional UKA with respect to tibial implant positioning, post-operative limb alignment, and JL restoration. This was translated in improved survival at mid-term follow-up favouring the robotic group.