Background. We have conducted a randomized trial with thyroxine (T4) in 200 infants <30 weeks’ gestation. T4 treatment was associated with better 5-year outcome in infants <29 weeks’ gestation, but with worse outcome in infants of 29 weeks. These effects could be related to low, respectively high free thyroxine (FT4) levels Methods. For each infant, the average FT4 of 5 scheduled measurements was calculated between day 3 and day 28. Infants of the placebo and the T4 group separately were divided in 2 groups. The placebo group consisted of a group of infants with average FT4 in the lowest quartile and a group in the upper 75%. The T4 group consisted of a group of infants with average FT4 in the upper quartile and a group in the lower 75%. Developmental outcome (mental/cognitive, motor, and neurologic) at 2 and 5.7 years was compared between high and low FT4 groups, and then compared separately for the T4 and placebo group. Results. In the placebo group, low FT4 was associated with worse outcome on all domains at both time points. After correction for confounding variables, mental and neurologic outcome remained significantly different at 2 years, and motor outcome at 5 years. In the T4 group, high FT4 was not associated with worse outcome, neither at 2 nor at 5 years. Conclusions. In untreated infants, low FT4 values during the first 4 weeks after birth in infants born at <30 weeks’ gestation are associated with worse neurodevelopmental outcome at 2 and 5 years. In T4-treated infants, high FT4 is not associated with worse outcome. Other factors than high FT4 concentrations must play a role in the worse outcome of the T4-treated group of 29 weeks’ gestational age.