The results of 45 revisions performed for fracture of the femoral stem in patients with total hip replacements were reviewed after an average follow-up time of 5 (3-7) years. We were able to examine 38 of the patients. Surgery was performed in a standardized way through a posterior approach and with a dorsal proximal window in the femur. In 7 cases the femoral stem only was changed, while in 31 both components had to be replaced. The femoral components implanted were of the revision type, and were cemented and individually selected in each case. To stabilize large acetabular defects a combination of an acetabular support ring and a homologous bone graft was used in 21 cases; bank bone was used in a total of 28 hips. During surgery 1 fracture and 4 defects of the femur occurred. Good intraoperative stability of the femoral component was attained in every case, allowing full weight-bearing during mobilization unless acetabular instability was present. There were no deep infections. One dislocation had to be treated by closed reduction. Clinical evaluation was performed according to Merle d'Aubigné. Radiological loosening was classified according to Gruen, DeLee and Charnley. Clinical results were good in 33 (87%) and were rated as poor in 5. In the latter group a good correlation between subjective and objective parameters was noted in 4 of the 5 patients. Two acetabular and two femoral components in 3 patients were classified as radiologically loose; in 1 of these the cup was not changed at revision.(ABSTRACT TRUNCATED AT 250 WORDS)