We reviewed the long-term results of core decompression for the treatment of nontraumatic osteonecrosis of the femoral head, performed in thirty-four patients (fifty-four hips) between January 1, 1981, and June 30, 1995. Twenty patients (59 percent) had bilateral involvement. The mean age of the patients at the time of presentation was thirty-eight years (range, twenty-two to eighty-three years). The presumed risk factors were use of corticosteroids (thirty-seven hips), excessive intake of alcohol (eight hips), and use of adrenocorticotropic hormone for the treatment of multiple sclerosis (two hips); the remaining seven hips had idiopathic osteonecrosis. According to a modification of the classification system of Ficat and Arlet in combination with the system of Steinberg et al., thirteen hips were stage I (normal radiographs) preoperatively; seven, stage IIA sclerotic; sixteen, stage IIA cystic or sclerocystic; ten, stage IIB (transitional stage, with a crescent sign); and eight, stage III (collapse). The mean duration of follow-up after the core decompression was 120 months (range, twenty-four to 196 months). The result was considered successful if the patient was asymptomatic, with no progression of the disease, and unsuccessful if there was radiographic failure (progression to stage III [collapse]) or clinical failure (the need for a subsequent operation), or both. The Kaplan-Meier product-limit method was used to estimate clinical and radiographic survival. Overall, twenty-six hips (48 percent) had a satisfactory clinical result and twenty (37 percent) survived according to radiographic criteria. Radiographic or clinical failure, or both, were seen in four of the thirteen stage-I hips, none of the seven stage-IIA sclerotic hips, thirteen of the sixteen stage-IIA cystic or sclerocystic hips, nine of the ten stage-IIB hips, and all eight stage-III hips. On the basis of the Cox proportional-hazards regression model, significant predictors of overall failure included an advanced preoperative radiographic stage (p < 0.0001), a shorter duration of symptoms (p < 0.05), and use of corticosteroids (p < 0.05). No association was found between age, gender, excessive intake of alcohol, or renal transplantation and the overall outcome. Two patients (two hips; 4 percent) had a postoperative complication. One patient had a fracture of the femoral neck, and the other had a hematoma. Our findings suggest that core decompression is a safe and effective procedure for the treatment of stage-I or stage-IIA sclerotic disease. These data also demonstrate the importance of differentiating between stage-IIA sclerotic disease and stage-IIA cystic or sclerocystic disease. We believe that core decompression has a limited role in the operative management of patients who have evidence of cystic changes in the femoral head on plain radiographs.