We prospectively studied the intermediate-term results of forty-nine revision total hip arthroplasties without cement that were performed because of aseptic loosening of a cemented femoral component in forty-five consecutive patients; the mean duration of follow-up was sixty-five months (range, forty-five to eighty-seven months). A curved, long-stem, titanium-alloy, non-circumferentially porous-coated femoral component was implanted in each hip. Preoperatively, a staging system was used to classify deficiencies of femoral bone stock according to the loss of cancellous or cortical bone in the metaphysis and diaphysis. Forty-one hips (84 per cent) had cortical or ectatic cavitary bone loss in the metaphysis. The mean Harris hip score significantly improved from 54 points preoperatively to 84 points at the time of the latest follow-up examination (p < 0.001). Twenty-seven patients (twenty-eight hips; 57 per cent) had at least two millimeters of subsidence of the femoral component during the first postoperative year. Eight patients (eight hips; 16 per cent) had no further progression of subsidence. Twenty-one patients (twenty-two hips; 45 per cent) had at least two millimeters of subsidence on two separate postoperative evaluations and therefore were considered to have progressive subsidence. Seventeen patients (nineteen hips; 39 per cent) had no measurable subsidence and were considered to have a stable femoral component. One of these seventeen patients had had a bilateral femoral revision and had progressive subsidence on one side. There was a positive trend for an association between subsidence and the degree of preoperative femoral bone deficiency (p = 0.10), but there was no association between subsidence and the fit of the prosthesis in the metaphysis and diaphysis or the fill of the canal of the femur (p > 0.50). There was no significant loss of bone in the hips with either a stable or a subsided femoral component (p > 0.50), and qualitative reconstitution of the cortex was noted in eleven (52 per cent) of the twenty-one most severely deficient (stage-III) femora. Survivorship analysis showed that, at seventy-two months, there was a 96 per cent chance of survival of the component (95 per cent confidence limits, 0.89 to 1.0) with revision as the end point but only a 37 per cent chance of survival (95 per cent confidence limits, 0.15 to 0.59) with revision or progressive subsidence as the end point.(ABSTRACT TRUNCATED AT 400 WORDS)