Abstract Our initial experience with mobile bearing medial compartment unicompartmental arthroplasty (UKA) is presented to highlight lessons that have been learned to avoid short-term failures. Consecutive cases of the Oxford medial UKA performed between February 2001 and April 2006 were reviewed to derive those cases that were revised to total knee arthroplasty (TKA). There were 545 patients available with mean age and body mass index of 65.0 and 30.1, respectively. At final follow-up, 32 patients were revised for lateral compartment arthritis, aseptic component loosening, persisting medial or anterior pain and dislocated meniscal bearing. Revisions were performed with primary unconstrained TKA implants with no stems or wedges required. Our results seem to reflect those seen in registries confirming an earlier higher revision rate and highlight the technical issues of overstuffing the compartment, inadequate cementation technique, and strict adherence to patient selection.