Abstract Previous scientific data suggest that beta-cell destruction in type 1 diabetes is mediated by an autoimmune process. This work was aimed at expanding existing knowledge of humoral autoimmunity by analysing antibodies against the intracellular part of the IA-2 protein (IA-2A) in 1200 patients with the disease, 750 siblings and more than 370 non-diabetic controls. IA-2A were present at the time of diagnosis in the overwhelming majority of patients with type 1 diabetes, and were associated with human leucocyte antigen (HLA) DR4 and DQB1*0302, but not with gender. Humoral autoimmunity was more marked in patients diagnosed when younger than 20 years of age than in older ones, but no noticeable association was observed between IA-2A and age under the age of 20 years. IA-2A in combination with antibodies to GAD65 (GADA) identified a higher proportion of patients younger than 15 years of age at the time of diagnosis than did islet cell antibodies (ICA) alone. The levels of IA-2A and the proportions of antibody-positive patients decreased with increasing duration of type 1 diabetes, although more than half of the patients still tested positive for IA-2A after 10 years of clinical disease. IA-2A, GADA, insulin autoantibodies (IAA) and ICA were detected with individual fluctuations in 8-14% of the siblings of children with type 1 diabetes monitored from the time of diagnosis of the proband, and the fluctuations were modified by HLA-defined genetic susceptibility, age of the siblings, family size and total number of detectable autoantibodies. IA-2A positivity detected at the time of diagnosis of the proband increased the risk of future disease in the siblings. The positive predictive value increased with increasing IA-2A levels, although individual risk assessment appeared to be a complex matter. In conclusion, IA-2 appears to be an important autoantigen in type 1 diabetes, since IA-2A is associated with the HLA haplotype that most strongly predisposes subjects to the disease and have the highest positive predictive value for future disease out of the four autoantibodies used for risk assessment purposes.