Ninety-six cup arthroplasties (eighty-three patients) were converted to total hip arthroplasties with cement between July 1970 and August 1982. Fifty-eight hips (fifty patients) were followed for at least ten years, or to failure after a shorter interval, and eight other hips (eight patients) had a subsequent operation because of a deep infection. Of the fifty-eight hips that were followed for at least ten years or to failure, nine (16 per cent) were revised because of aseptic loosening of the acetabular component, one (2 per cent) was revised because of a traumatic fracture of the femur, and none were revised because of loosening of the femoral component. Kaplan-Meier survivorship analysis with revision for any reason (including infection) as the end point showed a rate of survival of 92 +/- 6 per cent (average and 95 per cent confidence interval) at ten years and of 74 +/- 12 per cent at twenty years. When the hips in which an infection had occurred were excluded, and with revision because of aseptic loosening of the acetabular component as the end point, the rate of survival was 84 +/- 10 per cent at twenty years; no acetabular component was revised because of aseptic loosening in the first ten years. When the hips in which an infection had occurred were excluded, and with radiographic evidence of definite or probable loosening of the acetabular component, or aseptic loosening of the acetabular component necessitating revision, as the end point, the rate of survival was 91 +/- 6 per cent at ten years and 53 +/- 22 per cent at twenty years. The present study demonstrates the durability of total hip replacement with cement after the failure of a cup arthroplasty and further substantiates the excellent long-term clinical and radiographic results that can be obtained with insertion of a femoral component with cement.