The development of insulin-dependent diabetes mellitus (IDDM) is associated with circulating antibodies to a pancreatic islet protein of MW 64,000 (64 kDa), reported to be glutamic acid decarboxylase (GAD). To investigate the antigenic properties of the 64 kDa antigen/GAD in IDDM we employed fetal pig pro-islets, a convenient source of islet antigens and an alternative to adult human islets for transplantation. Pro-islets contained a 64 kDa protein precipitated by antibodies in the sera of 9/14 (64%) at-risk IDDM, 12/33 (36%) recent-onset IDDM and 4/12 (33%) established IDDM patients and in 1/18 (5%) healthy control subjects. In addition, a 38 kDa protein was co-precipitated with the 64 kDa protein by 6/14 (43%) at-risk IDDM, 5/33 (12%) recent-onset IDDM, 1/12 (8%) established IDDM patient sera and 1/18 (5%) control subject serum. Both 64 kDa and 38 kDa antigens were specific to pro-islets; neither was detected in fetal pig thyrocytes, hepatocytes or splenocytes. The majority of the 64 kDa protein was co-precipitated with GAD enzymatic activity from pro-islets by either IDDM sera or a sheep anti-GAD serum. Previously, we showed that peripheral blood mononuclear cells (PBMC) from over half the subjects defined as being at risk for IDDM proliferate in response to fetal pig pro-islets. Proliferative responses of PBMC from pre-diabetic subjects to a crude extract of pro-islets were therefore measured before and after depletion of GAD by adsorption of the extract against GAD-1 monoclonal antibody. In 5/10 at-risk subjects, T cell stimulation indices were greater than the control mean + 2 SD and decreased in four by > 50% after depletion of GAD. In summary, a 64 kDa protein with the properties of GAD is present in fetal pig pro-islets and is recognized by both antibodies and T cells in a significant proportion of subjects at risk for early IDDM. Some subjects with anti-64 kDa antibodies also have antibodies that co-precipitate a 38 kDa pro-islet protein. Prospective studies of T cell reactivity in at-risk IDDM subjects are required to delineate the role of GAD and other candidate autoantigens in the initiation and progression of beta cell destruction.