Insulin autoantibodies (IAA) are early sensitive markers of pre-diabetes in the young. The aim of this study was to assess whether, using IgG-specific measurement with a protein A/G assay, IAA are already present at birth, and whether this assay is suitable for early autoantibody screening. Cord blood and follow-up samples from offspring of parents with type 1 diabetes included in the BABYDIAB study were analyzed. Although insulin antibodies in cord blood from children of mothers with type 1 diabetes were readily detected and correlated well with levels in the maternal circulation, no insulin binding was detected in 247 cord blood samples from children of father probands. IgG IAA were detected at 2 yr in all 21 children who had multiple islet autoantibodies or who later developed type 1 diabetes, but were confirmed in only 6 of 58 with IAA by the conventional IAA assay in the absence of other islet autoantibodies. False positive IAAs in the conventional assay were often attributable to hemolysis. Hemolysis did not affect protein A/G IAA measurement, and results in whole capillary blood samples were comparable to those in corresponding serum samples (r2 = 0.99). These data show that IgG IAA appear early and after birth, and that the protein A/G IAA assay is sufficiently sensitive for early screening. The specificity of this assay requires further evaluation.