Evaluation of therapy for Stevens-Johnson syndrome was initiated as a retrospective analysis and then extended to a prospective series of patients treated with corticosteroids. This report extends the initial prospective study of patients with Stevens-Johnson syndrome treated with corticosteroids and evaluates the total series of 41 patients relative to outcome and the presumptive etiology. We propose that management of Stevens-Johnson syndrome requires corticosteroid therapy and that the survival of patients with Stevens-Johnson syndrome may depend on this therapy. No fatalities or adverse effects due to corticosteroids were noted. Stevens-Johnson syndrome due to a drug, a drug metabolite or viral infection may mimic a graft-versus-host reaction in which the patient rejects skin, mucous membrane, kidney or liver cells to which the drug, drug metabolite, or virus has bound. Corticosteroids suppress the inflammatory rejection until the activating agent has been eliminated.