As hemophilic arthropathy infrequently affects the hip joint, we performed a multicenter retrospective study to determine the results of hip arthroplasty in hemophilic patients. Thirty-four hip arthroplasties were performed in twenty-seven male patients at four major hemophilia centers from October 1972 through September 1990. Twenty-six patients had classic hemophilia and one had factor-IX deficiency. The mean age of the patients at the time of the operation was thirty-eight years (range, fifteen to seventy- three years). The mean duration of follow-up was eight years, with a minimum of two years for all patients who were still alive at the time of this review. Four patients were seropositive for the human immunodeficiency virus at the time of the operation, and sixteen patients were seropositive at the time of the most recent follow-up examination. Nine patients (33 per cent) died before the time of this review; seven had been seropositive for the human immunodeficiency virus. There were twenty-six total hip arthroplasties performed with cement, six total hip arthroplasties performed without cement, one total hip arthroplasty in which the femoral component was inserted with cement and the acetabular component was inserted without it (so-called hybrid arthroplasty), and one bipolar arthroplasty performed with cement. There were nu early infections after these thirty-four primary arthroplasties. There were three late infections around prostheses inserted with cement, and all led to a resection arthroplasty. Six (21 per cent) of the twenty-eight cemented femoral components and six (23 per cent) of the twenty-six cemented acetabular components were revised because of aseptic loosening. Of the twenty-four cemented femoral components for which radiographs were available and that were still in place at the time of this review or at the time of death, ten were definitely loose, two were probably loose, five were possibly loose, and seven had no evidence of loosening. Of the twenty-three cemented acetabular components for which radiographs were available and that were still in place at the time of this review, ten were definitely loose, seven were probably loose, three were possibly loose, and three were nut loose. None of the prostheses that had been inserted without cement were loose. There was a high rate of loosening of the cemented hip prostheses in these patients. There was also a high rate of mortality over-all and a high rate of late deep infection in the patients who were seropositive for the human immunodeficiency virus.