Three hundred and sixty-three knees (221 patients) that had been treated with total arthroplasty between 1975 and 1985 were divided into three groups on the basis of the preparation of the surface of the bone and the technique of the cement application. Group 1 (155 knees) was treated with irrigation of the bone surfaces with a syringe and manual packing of the bone cement. Group 2 (sixty-one knees) had high-volume, high-pressure lavage of the bone surfaces and manual packing of the cement. Group 3 (147 knees) had high-volume, high-pressure lavage and pressure injection of low-viscosity methylmethacrylate cement. Radiolucency was evaluated at each follow-up interval, and the findings were subjected to chi-square analysis and Kaplan-Meier survival analysis. Chi-square analysis of the data at one and three years indicated a significantly higher frequency of lines at the bone-cement interface in the femur and tibia in Group-1 knees compared with those in Groups 2 and 3. The survival curves showed increasing rates of radiolucency adjacent to the tibial components of the knees that had been prepared with lavage that was not high-volume and high-pressure and that had been prepared with finger-packing of cement (Group 1). This may partially explain the clinical problem of loosening of tibial components despite acceptable alignment. This study demonstrates that proper preparation of the cancellous bone and pressurization of the cement reduce the initial occurrence of a radiolucent line, which may have a positive effect on the ultimate failure of at least the tibial component.