Treatment of acute optic neuritis with steroids has been shown to hasten visual recovery without affecting the final degree of recovery. However, MRI-clinical studies indicate that patients with long optic nerve lesions, particularly those that involve the nerve within the optic canal, may have a worse prognosis for recovery of vision. Partly because such lesions could lead to swelling and subsequent ischemic optic nerve damage, steroids could have a selective beneficial effect on this subgroup of patients. The present randomized trial was designed to test this possibility. Sixty-six patients with acute optic neuritis received IV saline or IV methylprednisolone. The clinical, psychophysical, electrophysiologic, and MRI outcomes were assessed after 6 months. Patients with short lesions presented earlier than those with long lesions (involving three or more 5-mm-thick slices of any part of the optic nerve, as well as its intracanalicular portion), and lesion length was significantly less in patients presenting within a week of onset of symptoms. Lesions also tended to lengthen during follow-up in individual patients. Treatment did not limit lesion length in either the long or short lesion subgroup and had no significant effect on final visual outcome. We conclude that steroids do not improve visual outcome or lesion length in patients with acute optic neuritis.