Robotic assistance in knee arthroplasty has become increasingly popular due to improved accuracy of prosthetic implantation. However, literature on the mid-term outcomes is limited especially that of hand-held robotic-assisted devices. We present one of the longest follow-up series to date using this novel technology and discuss the learning curve for introducing robotic technology into our practice. The purpose of this single-surgeon study is to evaluate the survival, patient-reported outcomes and learning curve for handheld boundary-controlled robotic-assisted unicompartmental knee arthroplasties (HBRUKAs) at our hospital. This retrospective study evaluates 100 cases (94 Medial, 6 Lateral) performed by a single surgeon between October 2012 and July 2018. 52% were males, mean age was 64.5y (range 47.3y-85.2y) and mean BMI was 31.3 (range 21.8–43). Both inlay (40%) and onlay (60%) designs were implanted. Patients were followed up routinely at 1 and 5 years with Oxford Knee Scores (OKS) recorded. The learning curve was determined by tourniquet times. At a mean follow-up of 4.3 years (range 1.6y–7.3y), survivorship was 97%. There were three revisions: One case of aseptic loosening (1.5y), one case of deep-infection (3.8y) and one case of contralateral compartment osteoarthritis progression (5y). Mean 5-year OKS was 39.8. A 14.3% reduction in mean tourniquet times between the first 25 cases (105.5minutes) and subsequent cases (90.4minutes) was seen. This single-surgeon study showed good survivorship and patient-reported outcomes for HBRUKAs at our hospital. A learning curve of approximately 25 cases was shown, with significant decreases in tourniquet times with respect to increased surgeon experience