The unicompartimental knee prosthesis known as "Oxford" is a non constraint prosthesis, entrusting the whole of its stability to an intact ligamentary apparatus. Where the support surfaces of most prostheses remain limited, even punctiform, the originality of the Goodfellow prosthesis lies in the fact that the prosthetic condyle, whatever the flexion angle is, leans against a mobile prosthetic meniscus with spheric superior concavity of the same radius as the condylian radius, which increases considerably the prosthetic leaning surfaces and therefore lessens the pressure constraints. The superior surface, concave, of this prosthetic meniscus takes charge of the rolling, where the inferior plane surface realizes the gliding on the metallic tibial plate. The total conformity of the components minimizes the forces of friction.Between July 1988 and March 1993, 24 patients underwent the placing of UCP. Three patients died and 2 were lost of sight. 19 patients could be seen again or checked, corresponding of 21 operated knees. Two knees benefited from the start from UCP (medial and lateral) and 2 knees had a UCP in the first instance and then a second UCP in the compartment left safe primarily. For the 21 UCP, there are 16 medial and 3 lateral. Our mean drawback is of 3 years and 3 months, all the drawbacks being superior to 1 year and 4 months. The mean age is of 64 years. There were 17 female and 2 male patients. The mean weight is of nearly 80 kg (79,8) and nearly 52% of the operated patients have an important overweight (Body Mass Index superior to 30). Preoperative clinical analysis. It is based on a retrospective study of files using the quotation described by AUBRIOT for the «GUEPAR¼ group. This one establishes a gradation of four levels for each of the three criteria retained (Pain, Mobility, Instability), thus determining a global result imposed by the lowest level retained.For walking, other factors than just the state of the operated knee may intervene, this being the reason why it doesn't show in this chart. The GUEPAR group quantifies it with letters A, B, C, D.Concerning pain, all 21 knees were quoted as "Bad" in preoperative. Pain constitutes the decisive argument for the operative indication. In our series, only one knee had an average amplitude, all the others had a mobility superior to 89°. In 5 cases there was a flessum between 11 and 20° (penalizing of a level). Concerning walking and stability, they were taken into account, thanks to a precise questionnaire about the daily life acts. Concerning the walking perimeter, it was found as unlimited (A) in 1 case, superior to 500 m (B) in 2 cases, inferior to 500 m (C) in 17 cases and limited to home (D) in 1 case. The early after effects. At the end of the intervention, the knee is placed into a splint with limited flexion. As soon as the second day the patient is sat on the border of his bed. The first partial support at the third of the body weight is authorized between the fourth and the fifth day, when at the same time flexion exercises on electrical splint are started, as soon as the Redon draining is removed. The average hospitalization length was of a fortnight. Among secondary late complications and retakes, let us stop on meniscal luxations which constitute a specific complication of the Oxford arthroplasty. They concern 3 times the medial compartment and 4 lateral compartment. They happened in 1 case early, at D 22, in 3 cases within the 6 first months and in 3 cases after 2 years. They were treated : 3 times by reduction under general anæsthetic, no more ; 3 times changing the meniscusus for a meniscusus of superior size and once by placing a total prosthesis at the place of the UCP. The deteriorations of the opposed compartment not prosthesized occured in three cases. They were treated by unicompartmental additional arthroplasty in two cases and by total prosthesis in the third case. The clinical results on pain are very satisfactory as from the early check up onwards we have 17 successes (no pain 11 cases and occasional pains 6 cases) and as after 3 years and 5 months in average, we have 19 successes (no pain : 10 cases - occasional pain : 9 cases). At the maximal drawback, the mobility is quoted very good in 7 cases and good in 13 cases, mean in 1 case. At the latest check up, we note an excellent stability in 17 cases and good in 3 cases, that is to say 20 successes and 1 case of stability quoted as mean. At the latest check up we note 17 successes (A and B) and 4 relative failures (C) concerning the quality of walking.At the question «are you pleased with the intervention and would you advise it to a friend?¼ and with the nuance «very pleased¼ and «simply satisfied¼, we get 10 cases «very pleased¼, 8 cases «pleased¼ and 3 cases «moderately satisfied¼; only those 3 cases advise against the intervention. The radiological results are less satisfying as they show frequent imperfections : • for the 16 medial UCP : only 9 cases hypocorrected or normo axed, but 1 case strongly hypocorrected (residual varus of 7°) and 6 hypercorrected cases. • for the 5 lateral UCP : 3 normo-axed cases, 1 case strongly hypocorrected (residual valgus of 6°) and 1 case strongly hypercorrected (10° varus). • the failures due to rapid deterioration of the non prosthetized compartment occurred on hypercorrected knees. • on 21 knees, 14 borders of tibial plate were noticed, out of which 9 had no plate displacement and 5 had a slight displacement, at the origin of a small angular loss. • accumulations of cement on the tibial side, towards the back or in medial were noticed in 8 cases, which explains a slope of the tibial plate to the back inferior to 5° in 11 cases (should be of 7°). • 4 femoral components seem to be too posterior and one shows curved.In total, only 7 cases out of 21 were estimated with no peculiarities on the radiological point of view. It seems difficult to place a UCP well. The meniscal luxations are favored by an alignment rotational defect of the tibial plate, specially for the lateral UCP, the meniscus coming to hit the lip of the tibial plate during the lifting from a sitting position. For 5 of these luxations, we must recognize the existence of a ligamentary collateral laxity which should have altered the surgical indication either to an osteotomy, or to a total arthroplasty. Conclusions. Under the condition of respecting the absolute counter indications, of thoroughly evaluating the relative counter indications and of reducing at the best the defects linked to the surgical technique, the unicompartmental arthroplasty, including that of Oxford, gives good functional results after more than three years. In our series, the result on pain is constant if we exclude the cases with risk with ligamentary laxity and that of centered gonarthrosis at obese subject, that is to say 15 successes on 15 knees thus selected retrospectively. The gain on mobility is weak, of 5° in average. The result on stability is, as for pain, excellent, if we exclude the cases with risk, as we get then also 15 successes on 15 knees. Concerning the global result according to the quotation of Aubriot-Guepar, we note 14 successes and 1 relative failure. 4 knees were bad indications and should have benefited from a total arthroplasty or from an osteotomy.