To evaluate the emergence of diabetes-associated autoantibodies in young children and to assess whether such antibodies can be used as surrogate markers of type 1 diabetes in young subjects at increased genetic risk, we studied 180 initially unaffected siblings (92 boys and 88 girls) of children with newly diagnosed type 1 diabetes. All siblings were younger than 6 yr of age at the initial sampling, and they were monitored for the emergence of islet cell antibodies (ICA), insulin autoantibodies (IAA), glutamate decarboxylase antibodies (GADA), and IA-2 antibodies (IA-2A) up to the age of 6 yr and for progression to clinical type 1 diabetes up to the age of 10 yr. All 160 siblings with DNA samples available were typed for susceptible (DQB1*02 and *0302) and protective (DQB1*0301 and *0602-03) HLA DQB1 alleles. Twenty-two siblings (12.2%) tested positive for ICA in their first antibody-positive sample before the age of 6 yr, 13 (7.2%) tested positive for IAA, 15 (8.3%) tested positive for GADA, and 14 (7.8%) tested positive for IA-2A. There were 16 siblings (8.9%) who had 1 detectable autoantibody, 5 (2.8%) had 2, and 12 (6.7%) had 3 or more. In the group of 82 siblings with increased human leukocyte antigen-defined genetic susceptibility [DQB1*02/*0302, *0302/x (x = other than *02 or a protective allele), *02/y (y = other than *0302 or a protective allele)], 18 (22.0%) tested positive for ICA in their first antibody-positive sample, 10 (12.2%) tested positive for IAA, 14 (17.1%) tested positive for GADA, and 12 (14.6%) tested positive for IA-2A. One antibody was detectable in 6 siblings (7.3%), 2 were detectable in 5 (6.1%), and 3 or more were detectable in 12 (14.6%). Fifteen siblings (18.3%) presented with clinical type 1 diabetes before the age of 10 yr. All of the progressors showed increased human leukocyte antigen-defined genetic susceptibility. Thirteen of those 15 siblings, who presented with clinical type 1 diabetes before the age of 10 yr, had at least 2 antibodies detectable before the age of 6 yr (disease sensitivity, 87%; 95% confidence interval, 60-98%). Thirteen of the 17 siblings who tested positive for 2 or more autoantibodies before the age of 6 yr developed type 1 diabetes before the age of 10 yr (positive predictive value, 76%; 95% confidence interval, 50-93%). These observations suggest that disease-associated autoantibodies can well be used as surrogate markers of clinical type 1 diabetes in primary prevention trials targeting young subjects with increased genetic disease susceptibility.