Early implants for total knee replacement were fixed to bone with cement. No firm scientific reason has been given for the introduction of cementless knee replacement and the long-term survivorship of such implants has not shown any advantage over cemented forms. In a randomised, prospective study we have compared cemented and uncemented total knee replacement and report the results of 139 prostheses at five years. Outcome was assessed both clinically by independent examination using the Nottingham knee score and radiologically using the Knee Society scoring system. Independent statistical analysis of the data showed no significant difference between cemented and cementless fixation for pain, mobility or movement. There was no difference in the radiological alignment at five years, but there was a notable disparity in the radiolucent line score. With cemented fixation there was a significantly greater number of radiolucent lines on anteroposterior radiographs of the tibia and lateral radiographs of the femur. At five years, our clinical results would not support the use of the more expensive cementless fixation whereas the radiological results are of unknown significance. Longer follow-up will determine any changes in the results and conclusions.