The identification of disease-specific autoantibodies to the 65-kDa isoform of glutamate decarboxylase (GAD65Ab) epitopes in type 1 diabetes has been hampered by their conformational nature. Here, we compared two methods of GAD65Ab epitope analysis: GAD65/67 fusion proteins and competition assays using GAD65-specific recombinant fraction antigen binding (rFab). Sera from newly diagnosed type 1 diabetes patients (n=61) were studied using both approaches. Competition of GAD65 binding by an rFab to a specific epitope did not correlate with binding to the fusion protein that represented this epitope. Conversely, samples that bound to specific fusion proteins were not necessarily competed with rFab specific to determinants in the same region. We conclude that epitopes of different characteristics are detected by fusion proteins and by competition with rFab. Fusion proteins allow the definition of large epitope regions; however, some conformational GAD65Ab epitopes, especially those residing in the middle region, are destroyed or distorted in the fusion proteins. Competition studies using rFab allow the identification of conformational epitopes. However, monoclonal rFab may only reflect a limited proportion of the epitopes recognized by polyclonal sera. A combined analysis using both approaches may therefore be necessary to gain best understanding of autoantibody characteristics and affinity maturation.