The 20-year performance of metal-on-metal hip articulations has not been reported. Five McKee-Farrar total hip prostheses and one Sivash prosthesis were obtained at revision surgery after a mean implantation time of 21.3 years. A radiographic, histologic, implant, and wear analysis was performed on these total hip implants with cobalt-chrome metal-on-metal articulations. All cases were associated with femoral component loosening, but the bearing surfaces performed remarkably well. The worst case estimate of combined femoral and acetabular linear wear was 4.2 microns per year, about 25 times less than that typically seen with polyethylene. Metal particles and foreign-body inflammation were seen in all cases, but the volume of reactive tissue was small compared with what is generally seen at revision of hips with a polyethylene acetabular bearing. This may be due to a reduced particle burden or a decreased inflammatory reaction to particulate metal, or both. In addition to articular wear, other sources of metal particles included femoral neck impingement on the acetabular rim, stem burnishing, and corrosion. Prosthetic hip reconstructions can fail for many reasons, including suboptimal femoral stem and/or acetabular cup design and/or fixation. By today's standards, the McKee-Farrar and Sivash stem and acetabular component designs are suboptimal; however, after more than 20 years of use, the metal-on-metal bearing surfaces in these cases demonstrated low wear and do not appear to be the cause of failure. Recent advances in total hip arthroplasty, which include improved implant design, materials, manufacturing, and fixation, combined with a better understanding of the mechanisms of implant loosening and failure, suggest that the cobalt-chrome metal-on-metal bearing be reexamined as an alternative to polyethylene when exceptional durability is required.