Recently, the GM2-1 pancreatic islet ganglioside, proposed as a potential autoantigen in type I diabetes autoimmunity, has been biochemically characterized and found to be a novel ganglioside structure. In the present study, we aimed to determine whether an autoimmune response toward this novel islet molecule is 1) present in type I diabetes and is specifically directed against this molecule and not to gangliosides in general and 2) predictive of disease in high-risk subjects. To this end, the following patients have been studied: 1) 24 newly diagnosed type I diabetic subjects, 20 islet cell autoantibody (ICA)-negative first-degree relatives of type I diabetic subjects, and 25 age-matched normal control individuals; and 2) 31 prospectively evaluated ICA+ first-degree relatives of type I diabetic subjects who were followed for up to 10 years, during which 14 of them developed type I diabetes. A direct assay for autoantibodies to GM2-1 and to other pancreatic gangliosides (GM3, GD3, GD1a) was developed using an indirect immunoperoxidase technique performed directly on thin layer chromatography plates. Anti-GM2-1 autoantibodies (all belonging to the IgG class) were expressed in a high percentage of newly diagnosed type I diabetic subjects (71%), while no significant difference was found in the expression of antibodies directed against other pancreatic gangliosides (GM3, GD3, GD1a) among the different groups studied. Anti-GM2-1 autoantibodies were also present in ICA+ relatives (64%) (P < 0.001 vs. control subjects and ICA-relatives): in this group, life table analysis of progression to diabetes showed that anti-GM2-1 autoantibodies were significantly (P < 0.001) associated with disease, occurring in all relatives developing type I diabetes within 5 years and thus identifying a cohort of ICA+ subjects with markedly increased diabetes risk.