In a prospective, non-randomized study of 344 patients who had 392 primary total knee replacements with a Miller-Galante I prosthesis for the treatment of osteoarthrosis, the results of partial fixation with cement (insertion of the tibial and patellar components with cement and of the femoral component without cement) were compared with those of fixation without any cement. Of the 392 knees, 183 (163 patients) had fixation without cement (Group I) and 209 (181 patients), with and without cement (Group II). The average duration of follow-up was three years (range, two to five years). Nine patients died during the follow-up period, but no others were lost to follow-up. Analysis of the knee scores, range of motion of the knee, radiographs, and rates of complications revealed no differences between the outcomes in the two groups during the follow-up period. The rate of complications due to problems related to the extensor mechanism was high in both groups: a reoperation was performed in fifteen (8 per cent) of the knees that had had fixation without cement and in nineteen (9 per cent) of those that had had both types of fixation. Thirteen patients had additional operative treatment for recurrent patellar dislocations; twelve patients, for abnormal wear of the polyethylene of the patellar component: two patients, for avulsion of the patellar ligament from the tibia; and two patients, for unexplained pain in the knee. In addition, there were eight patellar fractures (two of which led to a reoperation) and three deep infections (all of which led to a reoperation).