Thirty-seven ankles in twenty-four patients were treated at our institution between July 1, 1974, and December 31, 1996, for atraumatic osteonecrosis of the talus. This group represents 2 per cent of the 1056 patients who were managed for osteonecrosis during this period. There were twenty-one women and three men, and their mean age was forty years (range, twenty-six to sixty-two years) at the time of the diagnosis. Thirteen (54 per cent) of the twenty-four patients had bilateral involvement. Sixteen patients (67 per cent) had a disease that affects the immune system, including systemic lupus erythematosus (thirteen patients), scleroderma (one), insulin-dependent diabetes mellitus (one), and multiple sclerosis (one). Four patients had a history of regular alcohol use, and four patients had a history of moderate smoking. One patient had a protein-S deficiency, one patient had had a renal transplant, and one patient had a history of asthma. Two patients had no identifiable risk factors for osteoarthrosis. Fifteen patients (63 per cent) had involvement of other large joints.The mean duration of symptoms before the patients were seen was 5.4 months (range, two months to two years). The mean ankle score at the time of presentation was 34 points (range, 2 to 75 points), according to the system of Mazur et al. A radiographic review revealed that, according to the system of Ficat and Arlet, eight ankles had stage-III or IV disease of the talus at presentation. The remaining twenty-nine ankles had stage-II disease. The osteonecrosis was seen in the posterolateral aspect of the talar dome (zones III and IV on the sagittal images and zones II, III, and IV on the coronal images) in twenty-two of the twenty-three ankles for which magnetic resonance images were available. The osteonecrosis was seen in the anteromedial aspect of the talar dome (zones I and II on the sagittal images and zone I on the coronal images) in the remaining ankle. Bone scans, which were available for eleven ankles, revealed increased uptake in the talus.All patients were initially managed non-operatively with restricted weight-bearing, an ankle-foot orthosis, and use of analgesics; two ankles responded to this regimen. Thirty-two ankles that remained severely symptomatic were treated with core decompression, which was useful in the treatment of precollapse (stage-II) disease. Twenty-nine of these ankles had a fair-to-excellent clinical outcome a mean of seven years (range, two to fifteen years) postoperatively; the remaining three ankles had an arthrodesis after the core decompression failed. Three ankles were treated initially with an arthrodesis for postcollapse (stage-III or IV) disease. All six of the ankles that had an arthrodesis fused, at a mean of seven months (range, five to nine months) postoperatively.When patients who have a history of osteonecrosis are seen because of pain in the ankle, the diagnosis of osteonecrosis of the talus should be considered. Early detection may allow the ankle to be treated non-operatively or with core decompression and thus reduce the need for arthrodesis. We also believe that when a patient has osteonecrosis of the talus, the hips should be screened with use of standard radiography or magnetic resonance imaging, or both.