The association between positive thyroid antibodies and an increased miscarriage rate in pregnancies after assisted reproduction technology (ART) remains controversial. We wanted to clarify this issue by performing a prospective cohort study in 234 women by systematically screening for thyroid peroxidase antibodies (TPO-Ab), serum TSH, and free T(4)(FT(4)) before the first ART cycle. Women with overt thyroid dysfunction were excluded. Fourteen percent of the cohort had positive TPO-Ab. Baseline characteristics [age, 33 +/- 5 yr; TSH, 1.6 (0.02-4.1) mU/liter; and FT(4), 12.2 (9.1-18) ng/liter] were comparable to those of the 86% of women without antibodies [age, 32 +/- 5 yr; TSH, 1.3 (0.05-3.6) mU/liter; and FT(4), 11.7 (9.5-16.5) ng/liter]. In the antibody-positive group, the pregnancy rate was 53% vs. 43% in the antibody-negative group, with an odds ratio of 0.67 [95% confidence interval (CI) (0.32-1.41); P = not significant]; however within the group that was pregnant, the miscarriage rate was 53% and 23%, respectively, with an odds ratio of 3.77 [95% CI (1.29-11.05); P = 0.016]. The age of the women was an independent risk factor for miscarriage, odds ratio 1.08 [95% CI (1.03-1.15); P = 0.005]. We conclude that women with positive TPO-Ab before the first ART cycle have a significantly increased risk for miscarriage.