We assessed a highly sensitive immunoradiometric thyrotropin (TSH) assay in screening thyroid dysfunction in 130 consecutive outpatients from a department of medicine and 224 patients from a municipal health centre. In addition to clinical examination, three routine tests were done: a thyroxine radioimmunoassay, an analogue-based free thyroxine assay and an immunoradiometric TSH assay. Triiodothyronine and the TRH test were done, if the findings were discrepant. Discrepancy existed in 24% of cases. The TSH assay had no false negative results (sensitivity 100%). Therefore TSH could screen all patients with thyroid dysfunction. Free thyroxine was the most specific assay (specificity 96%), but many subclinically or overtly hypothyroid patients would have been missed, if that assay had been used alone. We conclude that TSH(IRMA) is the best first-line measurement for thyroid dysfunction testing among outpatients. An abnormal TSH result alone is not diagnostic, but should be followed by the measurement of thyroxine or free thyroxine.