We compared results of five sensitive immunometric assays of serum thyrotropin (TSH) in controls and in different groups of patients with hyperthyroidism, untreated or treated; secondary hypothyroidism; nonthyroidal illness (NTI); or depression; or who were being treated with amiodarone. With most kits, measured TSH concentrations did not overlap between controls and hyperthyroid patients. In untreated secondary hypothyroidism TSH was not always undetectable. Patients with NTI and depression showed many low TSH values, and among these categories of patients, we observed large discrepancies among the kits. This lack of specificity at low concentration means that one cannot assess hyperthyroidism by TSH measurement alone, but it can be used as the first screening test. Similarly, TSH determination cannot be used alone in monitoring therapy (e.g., with carbimazole, thyroxin, amiodarone) to assess the presence of hyperthyroidism. Nonetheless, this assay plays a well-established role in hypothyroidism detection. Four of the five kits were found useful for clinical evaluation, the fifth less so.