The survivorship method of analysis has been used to compare the failure rate and overall success of 1,430 cemented primary total knee arthroplasties performed at The Hospital for Special Surgery over a 15-year period. There were 224 total condylar prostheses with a polyethylene tibia, 289 of the posterior stabilised type with an all polyethylene tibia, and 917 posterior stabilised with a metal-backed tibial component. There were 12 failures in the total condylar series, giving an average annual failure rate of 0.65% and a 15-year success rate of 90.56%. The posterior stabilised prosthesis with a polyethylene tibia showed an average annual failure rate of 0.27% and a 10-year success rate of 97.34%, and this prosthesis with a metal-backed tibial component gave an annual failure rate of 0.19% and a seven-year success rate of 98.75%. The overall survival rate was not influenced by sex or age, diagnosis or the percentage of ideal body weight. No metal-backed tibial components have yet needed revision for loosening. It seems that infection will be the major cause of failure.