We studied a large population (n = 2575) of unselected ambulatory persons older than 60 years to determine the prevalence of a low serum thyroid-stimulating hormone (TSH) level, ie, of less than 0.1 mU/L using a sensitive assay, a level suggestive of hyperthyroidism in younger adults. One hundred one persons (3.9%) had a low serum TSH level. About half of them (51/101) were taking thyroid hormone. Of the remainder, 44 were not hyperthyroid did not become so during up to 4 years of follow-up. Forty-one of the 44 euthyroid persons had a serum thyroxine level of less than 129 nmol/L; repeated testing showed a serum TSH level of more than 0.1 mU/L in the three euthyroid persons with a serum thyroxine level of more than 129 nmol/L. Only six were hyperthyroid or became so during the follow-up period; all had a serum thyroxine level of more than 129 nmol/L. Routine clinical examination was not a sensitive indicator of hyperthyroidism and did not permit discrimination from euthyroidism. A low value of serum TSH alone, while it had high sensitivity and specificity for hyperthyroidism, had a low positive predictive value (12%) for this diagnosis; addition of the thyroxine assay raised the predictive value fivefold to 67%. A low value of serum TSH is far more common in older persons than is hyperthyroidism. Low values in euthyroid persons are accompanied by a clearly normal serum T4 concentration (less than 129 nmol/L) or by a serum TSH level of more than 0.1 mU/L on repeated testing. We recommend measurement of the serum TSH thyroid concentration, using a sensitive assay, as the initial step in testing any older person for possible hyperthyroidism. Measurement of the serum T4 concentration or the free T4 index on the same sample would be needed only in the approximately 2% with a serum TSH level of less than 0.1 mU/L; alternatively, the TSH assay in these could be repeated at a later time.