From January 1979 to February 1982, 143 patients (seventy-nine women and sixty-four men) with 146 uninfected cemented total hip arthroplasties had revision cemented hip arthroplasty at The Hospital for Special Surgery for what was considered to be mechanical failure. The average age of the patients at primary arthroplasty was 56.1 years and at revision, 62.1 years. Loosening of the femoral component before revision correlated with varus positioning in 50 per cent of the hips, inadequate cement in 34 per cent, and a relatively young age in 16 per cent. The average age of the patients (fifteen hips) with a loose femoral component that had been placed in a neutral or valgus position with good cementing technique was 48.2 years at the time of primary arthroplasty. Loosening of the acetabular component was attributed to high placement of the cup in 41 per cent, inadequate bone in 18 per cent, a vertical orientation of the opening of the cup in 7 per cent, and poor cementing technique in 3 per cent. Complications associated with revision included perforation of the femoral cortex in 13 per cent, postoperative deep infection in 3.4 per cent, postoperative dislocation in 8.2 per cent, trochanteric complications in 6.2 per cent, and sciatic palsy in 0.7 per cent. Of the 139 hips that were followed for an average of 3.6 years (range, two to five years) after revision, the results were excellent in 59 per cent, good in 7 per cent, fair in 16 per cent, and poor in 18 per cent. After revision of the 139 hips, 29 per cent showed progressive radiolucencies; 18 per cent, femoral subsidence; and 9 per cent, acetabular migration. Definite mechanical failure after revision was identified in 15.8 per cent of the hips. These failures were due to loosening in 12.2 per cent of the hips, femoral fracture in 2.2 per cent, and disabling dislocation in 1.4 per cent. At the time of follow-up, twelve hips (8.6 per cent) had been revised a second time: six (4.3 per cent) for loosening of one or both components, three (2.2 per cent) for femoral fracture, and three (2.2 per cent) for infection. Mechanical failure and progressive radiolucencies were associated with poor quality of bone (p less than 0.001) and inadequate anatomical reconstruction (p less than 0.03).