Osteoarthrosis (OA) is a growing medical problem in western societies and the cost of the treatment has grown accordingly in the last years. Patients with OA often need to be operated on with arthroplasties and one important outcome measure for this type of surgery is the revision rate. The purpose of this study was to assess the results of knee arthroplasties performed at Akureyri University Hospital during 1983-2003, with special emphasis on revision rates, infections and other complications.Information was gathered from journals of the 457 patients who underwent knee arthroplasties during 1983-2003. Information about the operation and hospital stay was recorded as well as if patients later became the subject of revision. CRR (cumulative revision rate) uses survival statistics to estimate the risk of revision after primary operation and was calculated for patients with OA operated on with knee arthroplasty. Statistical informations were calculated in Microsoft Excel. Kaplan Mayer analysis was used to calculate the CRR and that was done in SPSS 11.5.560 primary operations were performed during the period, 515 total knee arthroplasties and 45 unicompartmental. 200 operations were performed on males and 360 on females. Mean ages for males was 70.8 years and for females 69.4 years. Revision rates varied depending on the type of implant. Twelve unicompartmental and 28 total knee arthroplasties became subject of revision. The PCA unicompartmental prosthesis most frequently needed revision, or in over 50% of cases. The CRR for the AGC total knee prosthesis was the lowest or around 3% at seven years, including revisions due to infections. Revisions due to infections were three in the period or 0.6% of all the total knee arthroplasties. Complications that substantially increase the risk of revision and/or are life-threatening were recorded in 1.8% of the operations at the time of discharge. Only one patient had pulmonary embolism (0.2%) and two patients (0.4%) had deep venous thrombosis.Our high revision rate for the PCA implant is consistent with what has been seen in other studies. This prosthesis was found to have mechanical problems and was withdrawn from the market. Our revision rate for the AGC implant as well as the rate of infections are low and the results are quite comparable to what has been found in Sweden by the Swedish Knee Arthroplasty Registry. The results of knee arthroplasties performed at Akureyri University Hospital, regarding revision rates, infections and complications, are fully comparable to other known results internationally.