Autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED) is caused by lack of functional products of the autoimmune regulator gene located on chromosome 21q22.3. The patients are at high risk of developing insulin-dependent (type 1) diabetes, but the positive predictive value of GAD65 or islet cell antibodies for type 1 diabetes is only 27%. Autoantibodies against the IA-2 tyrosine phosphatase-like protein (IA-2 ab) or insulin (IAA) have been suggested to be better markers for active ss-cell destruction. We studied these antibodies in sera from 60 Finnish patients with APECED, 12 of whom subsequently developed type 1 diabetes. Four (36%) of the 11 patients for whom we had prediabetic samples had IA-2 ab, and 4 (36%) had IAA. None of the 48 nondiabetics had IAA, and only 2 (4%) had IA-2 ab. Both had the antibodies for years without diabetes. Thus, IA-2 ab or IAA have a low sensitivity (36%), but high specificity (96% or 100%), with a positive predictive value of 67% for type 1 diabetes in patients with APECED. Data for human leukocyte antigen haplotypes were available for 59 of the patients, including 11 diabetics, and for 8 additional nondiabetic Finnish patients. No association between type 1 diabetes and high risk genotypes was seen. None of the 11 patients with type 1 diabetes, but 15 of the 56 (27%; P: < 0.05) nondiabetic patients and 24 of 93 (26%; P: < 0.05) of the control subjects had the DQB1*0602 allele, which is considered protective for type 1 diabetes. This is remarkable, as previously no positive or negative associations have been reported for any disease components of APECED with human leukocyte II antigens.