In this prospective study, 28 (52 percent) of 54 patients with systemic lupus erythematosus (SLE) had ischemic necrosis of bone in 93 sites. All but two of the patients had multiple sites of involvement, with hips, knees, and shoulders affected in decreasing order of frequency. Demographic, clinical, and laboratory features were comparable in patients with and without ischemic necrosis of bone except for cushingoid changes in 24 (86 percent) of the 28 patients with ischemic necrosis of bone versus four (15 percent) of the 26 patients without ischemic necrosis of bone (p less than 0.0001). The duration of steroid therapy, total cumulative steroid dose, and the mean daily prednisone dose for the first one, three, six, and 12 months of therapy were not significantly different between the two groups. Mean daily prednisone dose for the highest single month as well as the highest consecutive three, six, and 12 months of therapy was significantly higher in patients with ischemic necrosis of bone. The mean daily dose of prednisone for the highest month of therapy was greater than 40 mg per day in 93 percent and greater than 20 mg per day in all patients with ischemic necrosis of bone. In patients with ischemic necrosis of bone, there was a statistically significant correlation between higher mean prednisone dose at all time intervals and increased number of bony sites involved. A lower mean dose of prednisone was required to produce ischemic necrosis of bone in patients with Raynaud's phenomenon.