Hyperthyroidism has been considered to be underdiagnosed as a pathogenetic factor for atrial fibrillation, according to results of thyroliberin stimulation tests. We assessed, clinically and biochemically, thyroid function in 110 ambulatory patients with atrial fibrillation [mean age 64 (SD 11) years] from a group of consecutive cases referred to a specialist cardiology unit during one year. Patients finally categorized as euthyroid (n = 100) commonly presented with one or more symptoms or signs considered to be typical for thyrotoxic patients. Three patients (2.7%; 95% confidence interval, 0-7.5%) fulfilled criteria for hyperthyroidism, but only one was identified from clinical examination. The thyroliberin stimulation test (performed in all patients) identified another seven patients who might have been classified as hyperthyroid according to commonly used criteria for an abnormal thyrotropin response. However, none of these seven patients was judged as hyperthyroid after follow-up. Their thyrotropin concentrations were all above the detection limit for the immunoenzymometric assay. With few exceptions, they had thyroid hormone concentrations within reference limits calculated from the results for the 100 euthyroid subjects, and their concentration ratio for free 3,5,3'-triiodothyronine to free thyroxin, 0.31 (SD 0.05), was lower than that found in the euthyroid group [0.38 (SD 0.08), P less than 0.05]. We conclude that (a) criteria for evaluating results of the thyroliberin stimulation test should be revised, (b) this test is still useful in evaluating suspected cases of hyperthyroidism, for which results of current, improved methods for thyrotropin determination are equivocal, and (c) improved diagnostic methods should be used to characterize the relative importance of hyperthyroidism as a causal factor for atrial fibrillation.