Between September 1978 and December 1980, 28 revision procedures were carried out at the Wellesley Hospital, Toronto, for biomechanical failure of total hip arthroplasties. Twenty-six patients were personally reviewed and assigned a Harris hip rating. Seven patients had major complications intraoperatively or postoperatively; three of them required further surgery. Twenty-four patients had satisfactory results and none were worse. The mean Harris score preoperatively was 36.3. At a minimum follow-up of 1 year after revision, the mean Harris score rose to 79.9 (p less than 0.01). Revision of total hip arthroplasty is a difficult and demanding procedure but can produce satisfactory results.