In 157 new onset IDDM (104 men, 53 women, ages 10-39 yr) anti-thyroid peroxidase anti-bodies (anti-TPO) were assayed with a specific immunological test. Values greater than 100 U ml-1 were considered positive. Seventeen per cent of the patients were positive (32% of the women versus 10% of the men, p < 0.001). Eighty-five per cent of the anti-TPO + patients have a positive titre of islet-cell antibodies (ICA > or = 12 JDFU) versus 64% of the anti-TPO-patients (p < 0.05). When patients were subdivided in a young (10-25 yr) and an older age group (26-39 yr) this association was also true for ICA > or = 50 JDFU and valid for insulin autoantibodies (IAA) at low (> or = 0.7%) and high risk (> or = 1.5%) (p < 0.005) in the second group. The median of the TSH concentration was not different between anti-TPO+ and anti-TPO- when the group is considered as a whole. In the anti-TPO+ men (26-39 yr) TSH was however significantly greater (1.55 microU ml-1, range 0.74-8.5 versus 1.4 microU ml-1, range 0.21-3.5, p < 0.0001) when compared to the anti-TPO-men of the same age group. The haplotype HLA DQA1*0301-DQB1*0302 was more frequent in the anti-TPO+ (39%) than in the anti-TPO- (23%) patients (p < 0.02) for the age group 26-39 yr but not for the age group 10-25 yr. The other diabetes susceptibility haplotype DQA1*0501-DQB1*0201 was less frequent in anti-TPO+ patients. In conclusion we suggest that thyroid auto-immunity must be part of the initial screening of IDDM especially when patients are older at clinical onset of the disease.