The Oxford medial unicompartmental knee is designed fully congruent, with the purpose of maintaining a large contact-area throughout motion and minimize wear. No other study has investigated this design feature in-vivo. We aimed to evaluate if contact-loss was introduced between the articulating surfaces of the Oxford medial unicompartmental knee during bicycle- and step-cycle motion, and whether this correlated with essential implant parameters, such as polyethylene (PE) wear, knee-loadings, and clinical outcome. To study contact-loss, 15 patients (12 males, mean age 69 years) with an Oxford medial unicompartmental knee (7 cemented, mean follow-up 4.4 years) were examined with use of dynamic radiostereometry (RSA) (10 frames/s). PE wear was measured from static RSA and clinical outcome was evaluated with American Knee Society Score (AKSS) and Oxford Knee Score (OKS). Data on knee-loadings were acquired from the literature. Contact-loss was deteced in all patients during both exercises, and the trend of contact-loss correlated with the knee-loadings. Median contact-loss was 0.8 mm (95%PI: 0.3; 1.5) for bicycle motion and 0.3 mm (95%PI: 0.24; 0.35) for step-cycle motion, and did not correlate with the PE wear rate of mean 0.06 mm/year. Possible in-congruency was seen in three patients. Clinical outcome scores correlated with contact-loss during step-cycle motion. In conclusion, contact-loss was seen in all patients indicating a clinical tolerance during load. Contact-loss followed the knee-loadings, which could explain why no correlation was seen with PE wear, as an increase in load was acommadated by an increase in contact-area (contact-loss reduction). The size of contact-loss may reflect clinical outcome. © 2017 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 36:357-364, 2018.