The incidence of autoimmune diseases has shown a significant increase in developed countries during the last 40 years. The cause of this increase is still unknown, and reliable methods for the detection of individuals at risk of developing autoimmune disease are not available yet. To explore new methods for the diagnosis and monitoring of autoimmune disease, we have studied the repertoire of natural autoantibodies (NA) and its relationship with autoimmune disease using large arrays of defined antigens. NA are found in healthy humans and mice, apparently in the absence of immunization with their target antigens. We used knock-out mice to demonstrate that the repertoire of NA is influenced by factors not directly related to antigenic stimulation such as endogenous levels of histamine. By studying strains of mice known to differ in their susceptibility to autoimmune disease, we could conclude that the repertoire of NA reflects the susceptibility to develop autoimmune disease. The study of the human repertoire of NA required the development of bio-informatic tools to overcome the variation introduced by individual differences in the genetic background and immune history. We found that human NA are organized in clusters that can differentiate healthy subjects from patients with type 1 diabetes mellitus, type 2 diabetes mellitus or Behçet’s disease patients. The development of new tools to undertake large-scale NA analysis could also enhance our understanding of the immune system, and leave us in a better position to face the up-coming epidemics of autoimmune disorders.