The g-amino butyric acid (GABA) synthesising enzyme glutamic acid decarboxylase (GAD) is a common target of the humoral immune response in Type I (insulin-dependent) diabetes mellitus. Studies in first-degree relatives [1,2], twins [3] and school-children [4,5] have shown that GAD autoantibodies (GAD65-ab) can be present years or even decades before the clinical diagnosis of Type I diabetes. They have been used for prediction of diabetes development especially in combination with other diabetesrelated autoantibodies [6]. These GAD65-ab also ap- Diabetologia (2000) Abstract Aims/hypothesis. The aim of this study was to analyse the conformational and linear epitope profiles of glutamic acid decarboxylase antibody (GAD65-ab)-positive sera to find disease-specific epitope profiles and to study, whether GAD65-ab epitope recognition changes or spreads during the prediabetic period and, thus, can provide markers to differentiate early from later stages of progression to diabetes. Methods. Sera from subjects before (n = 21), at onset (n = 44), or at increased risk of Type I (insulin-dependent) diabetes mellitus (n = 20) and from patients with stiff-man syndrome (SMS, n = 18) or polyendocrine autoimmune syndrome (PAS, n = 21) were analysed for conformational and linear GAD65 epitope recognition by an immunohistochemical blocking test based on human monoclonal GAD65-ab (MICA 1±10) and western blotting of a GAD65 epitope-cDNA-library. Results. A redundant reactivity of many GAD65-ab positive sera to three major conformational (EP-1, EP-2, EP-3) and two dominant linear epitope clusters (amino acid 1±124 and 535±585) was observed in diabetes, polyendocrine autoimmune syndrome and stiff-man syndrome and no disease-specific epitopes or epitope-profiles were detected. Epitope recognition broadened with higher titres and with the vulnerability of patients to acquire additional autoimmune diseases apart from diabetes. Low GAD65-ab serum titres ( < 1200 arbitrary units) and EP-1 recognition in the absence of EP-2 binding characterised the early immune response. Changing epitope profiles combined stable recognition of EP-1 with gain or loss of reactivity to C-terminal epitopes during follow-up. Conclusion/interpretation. A maturing autoantibody response, which could spread from EP-1-recognition to other regions of GAD65, resulted in titre-related rather than disease-specific epitope profiles which were not sufficient to predict whether GAD65-ab positive subjects will progress to Type I diabetes, autoimmune polyendocrine syndrome or stiff-man syndrome. [Diabetologia (2000) 43: 210±217]