Between 6% and 20% of women of reproductive age have thyroid autoantibodies and there is evidence that this increases the risk of spontaneous abortion or preterm birth. A meta-analysis of published evidence (BMJ 2011;342: d2616; see also editorial, ibid: d 2260) has confirmed the increased risks and suggested that treatment with levothyroxine may reduce them. The meta-analysis included 31 studies (19 cohort, 12 case-control, 1216 women) about spontaneous abortion and five cohort studies (12566 women) about preterm birth, all including women with thyroid autoantibodies and apparently normal thyroid function. The presence of thyroid autoantibodies increased the risk of spontaneous abortion 3.9-fold in cohort studies and 1.8-fold in case-control studies. Among women with thyroid autoantibodies and recurrent spontaneous abortion or subfertility the increases in risk were 4.2-fold and 3.2-fold respectively. There was evidence of increased concentrations of thyroid stimulating hormone in women with thyroid autoantibodies and otherwise normal thyroid function. The studies on preterm birth showed a doubling of the risk in the presence of thyroid autoantibodies. Two small trials of levothyroxine in women with thyroid autoantibodies and normal thyroid function showed a halving of the risk of spontaneous abortion with treatment and one study showed a 70% reduction in the risk of preterm birth. Why thyroid autoantibodies have these effects remains uncertain. Two suggestions are that the women may have subtle hypothyroidism undetected by standard thyroid function tests or that they may …