Anti-neutrophil cytoplasmic autoantibody (ANCA)-associated vasculitis (AAV) is strongly associated with autoantibodies against myeloperoxidase (MPO) and proteinase 3 (PR3). No clear consensus has been reached on the pathogenicity of these autoantibodies. Animal models for MPO-ANCA, in vitro data suggesting pathogenicity of ANCA, and one case of a neonate showing symptoms of vasculitis after transplacental transfer of MPO, argue in favour of a pathogenic role for ANCA. On the other hand, the presence of natural MPO and PR3 autoantibodies in healthy individuals, lack of a strong correlation between ANCA titres and disease activity, and the occurrence of ANCA-negative AAV patients argue against pathogenicity of ANCA. Recent papers have drawn attention to the possibility of epitope specificity defining ANCA pathogenicity. Certain MPO epitopes were found to be specific for active disease, and others remained present during remission or were also present in healthy individuals. One linear epitope, aa447-459, was not only exclusive for active disease, but also detected in the total Ig fraction of ANCA-negative patients, reactivity being masked in serum by ceruloplasmin. So, not all ANCA seems to be equal, some could be pathogenic while others are not. For development of an autoimmune response, a specific ANCA repertoire is required, which may occur through intra-molecular epitope spreading in patients.