Revision of an acetabular component in a patient who has severe periacetabular bone loss is a complex problem, particularly when there is not enough bone stock to allow placement of an acetabular component near the normal anatomical hip center. A valuable option for revision in such a situation is placement of a hemispherical shell, fixed with screws and without cement, against the superior margin of the acetabular defect. The resulting hip center is more proximal than that seen following a typical primary total hip replacement. Forty-six hips in forty-four patients were treated consecutively, between July 1984 and February 1988, with a revision in which a hemispherical acetabular component was fixed with screws and without cement. All shells but one were placed with a so-called line-to-line fit. The procedures resulted in a so-called high hip center--that is, the center of rotation of the revised hip was located at least thirty-five millimeters proximal to the interteardrop line. The mean age of the patients at the time of the index procedure was fifty-two years (range, twenty-five to eighty-one years). The most common diagnosis for which the original arthroplasty was performed was osteoarthritis secondary to congenital hip dysplasia or dislocation (twenty-two hips). Thirty-four hips had had a high hip center before the index revision, and most patients had had a substantial limb-length discrepancy, with a mean of 1.6 centimeters of shortening on the side of the operation. In thirty-three hips, the femoral component was replaced as well, with a long-neck or calcar-replacement stem used when necessary to maintain or increase the length of the limb. Six patients (six hips) died before the minimum eight-year follow-up interval; none had had another revision or loosening of the revised acetabular component. Of the remaining patients, four (four hips) had the implant removed. One of them had a resection arthroplasty and one of them had a hip disarticulation because of infection after a subsequent femoral reoperation. Another had a hip disarticulation because of late infection. The fourth implant was removed because it had displaced into the pelvis at approximately six years; this was the only reoperation for aseptic loosening in the series. The remaining thirty-six hips (thirty-four patients) were followed for a mean of 10.4 years (range, 8.5 to 12.7 years). One acetabular component migrated medially and was scheduled for revision. No other acetabular component was loose or had been revised. The mean Harris hip score was 81 points (range, 56 to 100 points) at the time of the most recent follow-up. Despite the use of a high hip center, the prevalence of a positive Trendelenburg sign was reduced from 98 percent (forty-five of forty-six hips) preoperatively to 44 percent (sixteen of thirty-six hips) at the time of the most recent follow-up. The short limbs were lengthened a mean of seven millimeters (range, five millimeters of shortening to forty millimeters of lengthening). In this study of acetabular revisions with use of a high hip center in patients who had major periacetabular bone loss, mechanical failure occurred in 4 percent (two) of the forty-six hips in the entire series and in 6 percent (two) of the thirty-six hips in patients who were alive and still had the implant in place after a mean of 10.4 years of follow-up. The use of a high hip center did not adversely affect function of the abductor muscles, and the mean limb-length discrepancy was reduced by the femoral reconstruction.