BACKGROUND Nonunion after a periprosthetic femoral fracture associated with total hip arthroplasty occurs rarely. There is little information, to our knowledge, regarding the prevalence of this complication, its treatment, and the functional outcomes of treatment. The purpose of this study was to identify the patterns and frequency of nonunions of femoral fractures around total hip prostheses and to evaluate the results and problems associated with treatment of this complication in a consecutive series of patients. METHODS The study included twenty-three nonunions of periprosthetic femoral fractures in twenty-three patients with an average age of fifty-five years (range, twenty-two to eighty-five years) at the time of the initiation of treatment of the nonunion. Thirteen of the fractures occurred during or after a primary total hip arthroplasty, and ten occurred during or after a revision total hip arthroplasty. According to the classification system of Duncan and Masri, there were six B1 fractures (associated with a well fixed prosthesis), seven B2 fractures (associated with a loose stem), and ten B3 fractures (associated with very poor proximal bone). Ten patients were managed with revision to a long-stem prosthesis. Six patients had revision to a proximal femoral replacement prosthesis. A two-stage technique consisting of removal of the prosthesis and open reduction and internal fixation of the nonunion followed by reimplantation of the prosthesis was used in two patients. Two patients were managed initially with bone-grafting alone, and two patients were managed nonoperatively. One patient who had an infection at the site of the nonunion was managed definitively with resection arthroplasty. RESULTS The duration of clinical follow-up averaged 8.3 years (range, three months to twenty-three years), and that of radiographic surveillance averaged 7.0 years (range, eight months to seventeen years). Of the thirteen patients in whom an attempt to achieve union was made and for whom radiographs were available, nine eventually had bone-healing. Five of the twenty-three femora became infected and were treated with resection arthroplasty. Of the seventeen patients who had not had a resection arthroplasty for infection and for whom radiographs were available at the time of the most recent follow-up, eleven had a stable and well fixed implant and six had a loose implant as seen radiographically or had had a revision because of aseptic loosening. Seventeen patients had no or mild pain at the time of the most recent follow-up, but ten required two-handed support to walk. The overall complication rate was 52 percent (twelve of twenty-three patients). CONCLUSIONS Nonunion of a femoral fracture associated with a total hip prosthesis is an infrequent problem. Treatment is difficult, with a high rate of complications and relatively poor functional outcomes. The data from this series must be interpreted with caution, as patients were managed over a period of three decades and many did not have the advantage of modern techniques of revision hip arthroplasty. Prevention of nonunion by optimum treatment of the initial fracture is most important. Treatment of a femoral nonunion about a total hip implant should be implemented on the basis of the status of the fixation of the prosthesis and the quality of the surrounding bone.