e report the results of cancellous femoral impaction grafting with cement in revision hip arthroplasty in all patients from one centre who had undergone surgery more than five years previously. A total of 32 surgeons undertook femoral impaction grafting in 207 patients (226 hips). There were no deaths attributable to the revision surgery; 33 patients with 35 functioning hips died with less than five years' follow-up. One patient was lost to follow-up. Two hips (1%) developed early postoperative infection. Of the 12 stems which underwent a further surgical procedure for aseptic failure, ten were for femoral fracture and two for loosening.Survivorship with any further femoral operation as the endpoint was 90.5% (confidence intervals, 82 to 98) and using femoral reoperation for symptomatic aseptic loosening as the endpoint, the survivorship was 99.1% (confidence intervals, 96 to 100) at 10 to 11 years. As a consequence of the experience in this series, we have modified our technique with an increased use of longer stems with impacted allograft. Long stems are indicated when the host bone around the tip of a short stem is compromised, in patients with major loss of bone stock, or when a femoral fracture occurs.