We compared the effects of treatment of patients with prednisone or cyclophosphamide on a series of different types of autoantibodies. Levels of antiacetylcholine receptor (anti-AChR) antibodies and of antibodies to GM1 and GD1a gangliosides were measured in patients with a variety of neuromuscular disorders before and after treatment. Most patients had several autoantibodies present. We showed that prednisone treatment resulted in a reduction in titers of anti-AChR but not antiganglioside antibodies. Cyclophosphamide treatment produced a reduction of antiganglioside antibody titers. An intravenous and oral regimen was more effective than a single intravenous course of cyclophosphamide. We conclude that an immunosuppressive medication such as prednisone may reduce levels of some autoantibodies while producing no change in others, even in an individual patient. In addition, cyclophosphamide can suppress autoantibodies that prednisone does not. These differences in immunopharmacologic responses suggest that there are several distinct mechanisms of autoantibody production in humans. The utility of immunosuppressive medications in specific disease processes may be related in part to the mechanism of production of pathogenic antibodies.