Forty-five patients (forty-five hips) underwent repeat revisions of a total hip arthroplasty that had failed but was not associated with infection; seven of these patients had a third revision. The mean length of follow-up was approximately three years, and no patients were lost to follow-up. Twenty-eight of the forty-five patients had no or slight pain after the second revision, and thirty reported that their condition was improved. After the third revision, six patients had no or only slight pain, and five said that their condition was improved. On final roentgenographic examination, there was probable loosening (migration or subsidence of a component, lucency at the prosthesis-cement interface, fracture of the cement, or complete radiolucency at the bone-cement, or complete radiolucency at the bone-cement interface of more than one millimeter in at least one zone) of eight of the acetabular components and thirteen of the femoral components after the second revision and three acetabular components and one femoral component after the third. There was symptomatic loosening (moderate or severe pain and probable roentgenographic loosening) in six patients after the second revision and one after the third. Significant postoperative complications were noted in nineteen of the forty-five patients, and treatment was considered to be a failure in eleven hips after the second revision and in two after the third.