Objectivve To assess the prevalence and progression of subclinical hypothyroidism in women with type 2 diabetes.Design and Patients Cross-sectional and longitudinal observational assessment of thyroid function in 420 adult females with type 2 diabetes randomly selected from participants in the community-based Fremantle Diabetes Study.Measurements Serum TSH, antibodies to thyroperoxidase (anti-TPO) and serum free T4 were measured at baseline and after 5 years. Baseline glycated haemoglobin (HbA(1c)), serum glucose, serum total and high density lipoprotein (HDL) cholesterol, serum triglycerides and antibodies to glutamic acid decarboxylase (anti-GAD) were also used in analyses.Results After exclusion of patients with known thyroid disease or taking amiodarone or lithium at baseline, the prevalence of subclinical hypothyroidism (a raised serum TSH and normal serum free T4) was 8.6%. Subclinical hypothyroidism was associated with anti-TPO status and age, but there were no independent associations with serum cholesterol, history of coronary heart disease, HbA(1c) or hypoglycaemic therapy. In the subgroup of patients restudied after 5 years, none of those who had subclinical hypothyroidism at baseline had overt hypothyroidism regardless of anti-TPO status.Conclusions In women with type 2 diabetes without known thyroid disease, subclinical hypothyroidism is a common but incidental finding. The routine screening of thyroid function in type 2 diabetes is questionable.