A single injection of high-dose steroid (20 mg/kg) has been reported to induce necrotic lesions in the proximal metaphysis and diaphysis of the rabbit femur. In the rabbit osteonecrosis (ON) model induced by two-dose horse serum injections, contrast-enhanced magnetic resonance imaging (MRI) and T2*-weighted dynamic MRI have been reported to detect necrotic lesions at 3 days after the second serum injection sensitively. The purpose of the present study was to determine whether contrast-enhanced MRI and T2*-weighted dynamic MRI could detect early development of necrotic lesions in the rabbit proximal femora after a single high-dose steroid injection and compare MRI features of the two types of nontraumatic rabbit ON models. We performed nonenhanced MRI, contrast-enhanced MRI and T2*-weighted dynamic MRI of bilateral proximal femora 3 days (10 femora), 1 week (10 femora), 3 weeks (10 femora), 6 weeks (18 femora) and 9 weeks (18 femora) after a single 20 mg/kg steroid injection. Femoral signal intensity of each T2*-weighted dynamic MRI was measured from a 1-cm(2) region of interest in the proximal metaphysis and diaphysis. As a control, MRI was performed in untreated animals (six femora). Histologically, no necrotic lesions were observed in the proximal femora at 3 days and 1 week. Bone marrow necrosis was observed in four (40%) femora at 3 weeks, two (11.1%) femora at 6 weeks and six (33.3%) femora at 9 weeks. Bone marrow lesion completely replaced by granulation tissue was observed in one femur at 6 weeks and one femur at 9 weeks. Histologic evidence of repair tissue surrounding bone marrow necrosis was seen after 6 weeks. Average lesion area including repair tissue was 4.40 mm(2) (range, 0.32 to 20.2 mm(2)). At 9 weeks, contrast-enhanced MRI could detect four (66.7%) femora with bone marrow necrosis of more than 4 mm(2) in the lesion area, while T2*-weighted dynamic images showed a finding of complete ischemia in only one of these four femora. In conclusion, neither contrast-enhanced MRI nor T2*-weighted dynamic MRI could detect early development of necrotic lesions in the single-dose steroid ON model. These results indicated that development of necrotic lesions in the single-dose steroid ON model was not accompanied by as diffuse a femoral hemodynamic change as the two-dose horse serum ON model.