From June 1978 to December 1987, 106 revision total hip arthroplasties (THA) for acetabular salvage were performed using a bipolar device. Bone grafting was necessary in most of the patients. Five patients were unavailable for follow-up examination, leaving 96 patients (101 hips), who were followed for an average of two years 11 months (range, two months to 8.5 years). Excellent or good results were obtained in 43 patients; fair results were achieved in 20 patients. Of the nine patients with poor results, five demonstrated roentgenographic evidence of subluxation, and four showed no roentgenographic changes that could explain their persistent pain. Twenty-nine of these bipolar revisions failed and required reoperation. Fourteen of these failures were revised using fixed, cementless devices in conjunction with bone grafting. Ten patients developed deep-wound infection. Nine were treated successfully; the tenth patient died of overwhelming sepsis, her case complicated by multiple infected joint arthroplasties. While the results of revision THA in the present series are not as satisfactory as those reported by others who used fixed cemented or fixed cementless acetabular components, they are superior to the results obtained with excisional arthroplasty, the only alternative in many of these cases.