We reviewed the results an average of fifty months (range, twenty-four to 120 months) after the use of thirty-five allografts in thirty patients during primary or revision total knee replacement. Twenty-nine femoral-head allografts, five distal femoral allografts, and one proximal tibial allograft were used in conjunction with a long-stemmed implant to reconstruct large osseous defects. The patients were evaluated clinically, radiographically, and subjectively (with use of a questionnaire). Twenty-six (87 per cent) of the thirty patients had a good or excellent clinical result, and no revisions were necessary. As none of the patients had collapse of the graft, subsidence of the implant, or revision, we believe that the outcome of treatment with a femoral-head allograft, particularly in association with a component inserted with cement, is excellent. Four non-porous-coated components were placed without cement on structural allografts. Radiographically, three of those components subsided, but none of the three needed revision and two were associated with a good clinical result. Our current practice is to cement components in all arthroplasties involving grafting. Our findings suggest that the use of a stemmed component reduces the stress on the allograft, host bone, and fixation interface. In addition, such a component contributes to the longevity of a total knee replacement associated with a bone graft. Additional studies with long-term follow-up are necessary to confirm this outcome.