Between 1980 and 1986, ninety-three femoral revision arthroplasties without cement were performed on ninety-one patients who had symptomatic loosening of a stemmed femoral component of a total hip prosthesis. Eighty-one of these patients (eighty-three hips) were followed for at least five years (range, five to thirteen years; mean, nine years). Forty-three hips also had a revision of a cemented acetabular component with use of a hemispherical, porous-coated implant inserted without cement. The results are presented in a traditional manner in terms of implant survivorship, radiographic evaluation for stability of the implant, and standardized hip scores. Changes due to the treatment are presented as outcomes in terms of the physician's criteria for the success of the operation, the patient's criteria for the success of the operation, and the patient's economic status. No hips that needed additional operations were excluded, and we reported the outcome regardless of any interim operative procedures (including re-revisions). At the latest follow-up examination, nineteen (23 per cent) of the eighty-three hips had had an additional operation, including seventeen hips (20 per cent) for which re-revision had been necessary. In ten (12 per cent) of the eighty-three hips, the re-revision had involved an index implant. The rate of re-revision of the index femoral component was 10 per cent and the rate of mechanical loosening of that component was 11 per cent. The rate of re-revision of the index acetabular component was 7 per cent and the rate of mechanical loosening of that component was 11 per cent.(ABSTRACT TRUNCATED AT 250 WORDS)