SUMMARYThe serum T3 assay has been regarded as the most sensitive single test for hyperthyroidism although impaired conversion of T4 to T3 in non‐thyroidal illness (NTI) might decrease its diagnostic sensitivity. The present report gives experience from the T3 assay in middle‐aged females under conditions similar to those in a general health survey. The assays were performed during two periods with an interval of six years.In 1974–75 we studied a representative sample (n= 1283) of women of ages 44, 52, 56, 60 and 66 years in Göteborg, Sweden. Individuals with serum T3 concentration>mean + 2·5 SD were selected for a follow‐up study (n= 21). Of 16 individuals with no previous thyroid disease and no present treatment with thyroid hormones or oestrogens, 14 were subjected to a TRH‐stimulation test giving a normal TSH response in 10 cases having T3 concentrations up to mean + 3·5 SD. Four women with serum T3 concentration ≥ mean + 3·5 SD had previously unrecognized autonomous function thyroid function, of whom two developed hyperthyroidism after two years.The original population sample was reinvestigated after six years in 1980‐81 (n= 1138) together with an additional sample of women giving a total sample of 1422 women of ages 26, 38, 50, 58, 62, 66 and 72 years. Of the females studied in 1974–75 eight had developed hyperthyroidism between the two studies; three of these had raised serum T3 at the investigation in 1974–75. No case of hyperthyroidism had been missed by the T3 assay in the 1974–75 study. Of individuals with serum T3≥mean + 2·5 SD selected for a follow‐up (n= 29) at least five were found to have previously unrecognized thyroid autonomy.We found a raised serum T3 to be associated with hyperthyroid (n= 2) and euthyroid GRAVES' disease, autonomously functioning thyroid adenoma(s), possible painless subacute thyroiditis, possible thyrotoxicosis factitia, diminished thyroid reserve and thyroid substitution therapy.Since we did not correct for variations in protein binders a number of individuals with oestrogen‐associated rise of total T3 were identified, but the rises were usually modest. The incidence of hyperthyroidism for the original sample as calculated for the six‐year period was 1.3 cases/1000 women/year whereas the prevalence by 1980‐81 was 13 cases/1000 women.We conclude that a single T3 determination will in most cases be sufficient for the diagnosis of clinical hyperthyroidism in a non‐hospitalized population. Our study indicates some of the problems associated with the follow‐up of a general screening for hyperthyroidism, the possible