The smaller form of the autoantigen glutamic acid decarboxylase, GAD65 (formerly the 64,000 Mr autoantigen), is a major target of humoral autoimmunity in type I diabetes. Human autoantisera have been used extensively to characterize the GAD65 antigen in both rat and human islets, but the protein has escaped detection in mouse islets. We have now analyzed the expression of GAD65 and GAD67, the larger glutamic acid decarboxylase protein, in human, rat, and mouse islets of Langerhans and brain, using human monoclonal islet cell autoantibodies, human autoantisera, and experimentally raised antibodies to glutamic acid decarboxylase. Human monoclonal autoantibodies and experimentally raised antibodies reacted with mouse GAD65 produced in a baculovirus expression system by Western blotting and immunoprecipitation and with GAD65 in mouse brain by immunohistochemistry but failed to detect GAD65 in mouse islets by the latter two methods. However, analysis of mouse islets by Western blotting technique, using the most sensitive experimentally raised antibody, showed that mouse islets express both GAD65 and GAD67 but at levels that are severalfold lower than those in mouse brain or in human and rat islets. Furthermore, both human and rat islets predominantly express GAD65, whereas GAD67 is the major glutamic acid decarboxylase protein in mouse islets. Human islets are significantly distinct from mouse and rat islets and from brain because they only express GAD65, which is consistent with the predominant role of this form as a target of autoantibodies associated with β-cell destruction in humans. Human as well as rat islet GAD65 are found in both membrane-bound and soluble forms. The low level of glutamic acid decarboxylase expression in mouse islets compared with human and rat islets is likely to have implications for both the development of tolerance to glutamic acid decarboxylase as well as the homing of glutamic acid decarboxylase-specific lymphocytes to the mouse β-cell. In this context, the results suggest 1) that the mouse is ideal for studies of the consequences of an expression of high levels of glutamic acid decarboxylase in the β-cell from a transgene and 2) that the rat may be better suited than the mouse for development of nontransgenic animal models of glutamic acid decarboxylase autoimmunity by immunization.