Total hip replacement was performed in either one or two stages in thirty-three hips with active sepsis. The sepsis had followed hemiarthroplasty in six hips, open reduction with internal fixation of a fracture in eight, cup arthroplasty in one, and total hip replacement in eight hips within six years prior to the second total hip replacement. Ten additional patients had total hip replacement following destruction of the hip joint by hematogenous sepsis in nine and by infection following a shrapnel wound in one. Of these thirty-three patients, twenty-three (70 per cent) reveal no signs of infection at three to nine years after prosthetic replacement. Of the remaining ten in whom an infection developed, six had definite recurrences of the original infection, three were infected with organisms different from the original one, and one was either a local recurrence or reseeding from a persistent pyelonephritis. The success rate when the original organism was gram-positive was 78 per cent, including two of three total hip replacements done in the presence of active infection with Staphylococcus epidermidis. The success with gram-negative organisms, however, was only 58 per cent. The prosthetic failure rate was highest in patients who had had a previous infection about a total hip replacement (37 per cent) and in patients who had had a previous infection but no prior prosthetic or internal fixation devices (37 per cent). The lowest prosthetic failure rates were in patients with an infected hemiarthroplasty (16 per cent), an infection around an internal fixation device (25 per cent), or an infected cup arthroplasty. A complete and differential blood-cell count, erythrocyte sedimentation rate, aspiration arthrogram, and radiographs did not effectively predict success or failure. For gram-positive infections, the success rates were similar following either a one or a two-stage procedure. We found that the success rates could be improved by a repeat course of parenteral antibiotics after the total hip replacement even if all preoperative and intraoperative studies failed to identify an infection. Patients with a successful total hip replacement achieved much better functional results than those who had to have a Girdlestone procedure. However, all patients must be carefully assessed prior to reimplantation of a prosthesis because of the high failure rate, especially with gram-negative organisms (Pseudomonas having the gravest prognosis), even when the procedure is done in two stages.(ABSTRACT TRUNCATED AT 400 WORDS)