A fundamental goal in understanding the mechanisms of autoimmune disease is the characterization of autoantigens that are targeted by autoreactive antibodies and T cells. Unfortunately, the identification of autoantigens is a difficult problem. We have begun to explore a novel route to the discovery of autoantibody/autoantigen pairs that involves comparative screening of combinatorial libraries of unnatural, synthetic molecules for compounds that bind antibodies present at much higher levels in the serum of individuals with a given autoimmune disease than in the serum of control individuals. We have shown that this approach can yield "antigen surrogates" capable of capturing disease-specific autoantibodies from serum. In this report, we demonstrate that the synthetic antigen surrogates can be used to affinity purify the autoantibodies from serum and that these antibodies can then be used to identify their cognate autoantigen in an appropriate tissue lysate. Specifically, we report the discovery of a peptoid able to bind autoantibodies present in about one-third of nonobese diabetic (NOD) mice. The peptoid-binding autoantibodies were highly enriched through peptoid affinity chromatography and employed to probe mouse pancreatic and brain lysates. This resulted in identification of murine GAD65 as the native autoantigen. GAD65 is a known humoral autoantigen in human type 1 diabetes mellitus (T1DM), but its existence in mice had been controversial. This study demonstrates the potential of this chemical approach for the unbiased identification of autoantigen/autoantibody complexes.