We studied the cases of twenty patients who had had an ipsilateral total knee arthroplasty or a contralateral total hip arthroplasty, or both, long after one hip had been fused in an acceptable position. Between 1972 and 1986, we performed twenty-one total joint arthroplasties (on thirteen hips and eight knees) and followed two additional patients (one hip and one knee) in whom the operation had been performed elsewhere. The average age of the patients at the time of arthroplasty was fifty-seven years (range, thirty-one to eighty-one years), and the average time from arthrodesis to arthroplasty was thirty-two years (range, eleven to fifty-four years). The results of eighteen of the twenty-three arthroplasties were evaluated at an average of seven years and nine months postoperatively. Four of the remaining five patients, who were followed for an average of eight years, died of a cause that was unrelated to the arthroplasty. After the hip arthroplasty, five hips were rated excellent; five, good; one, fair; and three, poor. Each hip that had a poor result was revised twice for mechanical loosening. Three hips for which the result was not considered poor had progressive radiolucency. After the knee arthroplasty, three knees were rated excellent; four, good; one, fair; and one, poor (because of infection). Seven knees were manipulated a total of fifteen times. Only one patient had progressive symptomatic radiolucency, nine years after the insertion of a posterior stabilized prosthesis. Clinically important ligamentous instability was not encountered.(ABSTRACT TRUNCATED AT 250 WORDS)