Abstract Study question Does levothyroxine treatment increase live birth rates in thyroid peroxidase antibodies (TPO-Ab) positive women with recurrent pregnancy loss (RPL) and normal thyroid function? Summary answer Levothyroxine treatment did not result in higher live birth rates in women with RPL positive for TPO-Ab and a normal thyroid function, compared to placebo. What is known already Women positive for TPO-Ab have a higher risk of RPL. Levothyroxine supplementation has been proposed to reduce the risk of pregnancy complications. Evidence is limited whether levothyroxine treatment improves pregnancy outcomes in TPO-Ab positive women with RPL. Study design, size, duration We conducted an international double-blind trial. Between January 2013 and September 2019, we randomly assigned 187 women to receive levothyroxine (94 women) or placebo (93 women). Before conception, women were randomly assigned in a 1:1 ratio to levothyroxine or placebo orally once daily. The daily dose of levothyroxine was based on preconception TSH concentration and ranged from 0.5 to 1.0 µg/kg bodyweight. Levothyroxine or placebo was continued until the end of pregnancy. Participants/materials, setting, methods Patients with two or more pregnancy losses underwent diagnostic testing for RPL. Women with a normal TSH level and positive for TPO-Ab were eligible for the study. Antiphospholipid syndrome was an exclusion criterion. The primary outcome was live birth, defined as the birth of a living child beyond 24 weeks of gestation. Secondary outcomes included ongoing pregnancy, pregnancy loss, preterm delivery, adverse events and time to conception leading to live birth. Main results and the role of chance Live birth occurred in 47 women (50%) in the levothyroxine group and in 45 women (48%) in the placebo group (risk ratio, 1.03; 95% confidence interval [CI], 0.77 to 1.38; absolute risk difference 1.6%; 95% CI, -12.7% to 15.9%). There were no significant differences in any secondary outcomes. In both groups seven adverse events were reported, none of them directly related to the study procedure. Limitations, reasons for caution The slow recruitment forced us to stop the trial prematurely. With this sample size we are not able to fully account for dropouts and the estimates have wide 95% confidence intervals. Detection of a difference of 5% in live birth rate would have required inclusion of more than 3000 women. Wider implications of the findings Routine use of levothyroxine in women with RPL and normal thyroid function positive for TPO-Ab is not recommended. Our results are in line with previous trials in TPO-Ab positive women with a history of infertility or pregnancy loss or undergoing in vitro fertilization. Trial registration number Netherlands Trial Register (NTR3364).