One hundred forty-nine medial prostheses were implanted in 140 patients between 1988 and 1996. After a mean of 67 months 28 patients had died, without the need for revision. Seventeen prostheses were lost to follow-up. Revision surgery using a total knee prosthesis was performed in 16 cases. In four others, a lateral prosthesis was implanted subsequently to a medial one. One of these four was revised to a total knee prosthesis 6 years later. In another four cases, late complications of the meniscal bearing were treated with replacement of this bearing. The surviving prostheses were seen back after a mean of 126 months. The cumulative survival rate at 10 years was 82% for the whole population and 84% when knees with a previous high tibial osteotomy were excluded. Since these results compare poorly to the survival of total knee arthroplasty, this prosthesis is not the first-choice implant. Because it preserves a maximum of bone stock and is revised to a total prosthesis almost without difficulty, it is the first-choice implant for medial unicompartmental osteoarthritis in patients younger than 65. Further research is mandatory to confirm that this prosthesis very rarely needs revision in patients older than 75. It should not be used in osteotomized knees.