One hundred thirty-nine cemented and 132 cementless Miller-Galante total knee prostheses were followed between three and six years (average, 43-44 months). The fixation technique was based on patient age, bone quality, and ability to delay full-weight bearing. Clinical follow-up studies were possible on 116 cemented knees. Fifteen knees were lost because of death before the three-year follow-up study, and eight knees required component removal. One hundred twenty-three cementless knees were available for clinical follow-up studies; there were three deaths, and six failures required component removal. No cemented failure was due to fixation, and three cementless failures were due to lack of tibial ingrowth in two and pain of undetermined etiology in one. Preoperative knee scores were slightly significant with cemented knees averaging 48 points and cementless knees averaging 52 points. A similar significant difference was maintained at the final follow-up study. No significant differences were noted for pain, limp, or support scores. Average range of motion was similar in the two groups. Radiolucent lines about the femoral component were rare. Cementless tibial radiolucencies were partial in up to 20% of examined zones, and complete tibial tray radiolucency was seen in only three patients. No correlation between radiolucency and knee scores was seen.