The GAD65 epitope immunoglobulin binding pattern in cord blood of children (n=37), who later developed type 1 diabetes at 3.2-14.9 years of age, was analyzed. First, the binding at diagnosis was compared with that in the cord blood serum. The next comparison was between the cord blood serum and the mothers' serum taken at delivery. Basal GAD65 binding levels were determined in Protein A Sepharose-based radiobinding assays with (35)S-labeled human and rat GAD65, rat GAD67 and GAD65/67 fusion proteins representing N-terminal (N), middle (M) and C-terminal (C) epitopes. In the first comparison, 28/37 children had GAD65 binding above 2.44 relative units (RU) (upper three quartiles), representing a marked increase from birth in the binding to human GAD65 (p<0.0001), rat GAD65 (p<0.0001), N- (p=0.04), M- (p<0.0001), C- (p=0.001), and M + C-epitopes (p<0.0001), but not to rat GAD67. At birth, 9/37 had GAD65 binding above 1.56 RU (upper quartile) demonstrating that their binding of human (35)S-GAD65 was higher in cord blood than in the mother (p=0.008). Higher cord blood binding was also observed for the N- (p=0.02) terminal epitope but not for rat GAD65, rat GAD67, and the remaining epitopes. These data suggest that differences in the epitope GAD65 binding between mother and child at birth are limited. In contrast, the epitope pattern at diagnosis differed from that at birth, supporting the view that disease-associated epitopes develop between birth and diagnosis.