We have found poor mid-term results in a multisurgeon series of 94 Johnson-Elloy (Accord) total knee replacements. A total of 27 knees (29%) has required revision, in 26 for aseptic loosening. Only 18 (19%) remain in situ, and these give poor function, are painful and most show radiological evidence of early failure. At 12 to 13 years the survival rate is 43% (confidence interval 29 to 57) with failure requiring revision as the endpoint. Proximal migration of the femoral component is associated with considerable loss of bone stock. We believe that all patients who have this implant should be recalled for regular review in order to anticipate this problem.