Subclinical hyperthyroidism, defined as serum thyroid hormone levels in the reference range with low serum TSH concentration, is a well recognized clinical entity, but little information is available concerning the prevalence of the disorders that produce it. We conducted a 24-month retrospective survey of subclinical hyperthyroidism patients referred to a university hospital nuclear medicine service for diagnostic studies. Twenty-four consecutive patients were evaluated (22 outpatients and 2 inpatients). All patients had highly sensitive TSH determination, thyroid hormone levels, radioiodine uptake and scan (except for 2 postpartum women), and, selectively, TSH-receptor antibody (TRAb), serum thyroglobulin, antithyroid antibodies, T3-suppression test, and erythrocyte sedimentation rate. A TSH value of about 0.1 microIU/mL was used as the cutoff. Only one patient in the group had a nonthyroidal disorder. In 14 patients (61%) subclinical hyperthyroidism was self-limited, due to silent thyroiditis (5 patients), iodine-induced hyperthyroidism (3 patients), postpartum thyroiditis (2 patients), subacute thyroiditis (2 patients), and probable hemorrhage into a functioning nodule (2 patients). Of the non-self-limited disorders (39%), Graves' disease accounted for 6 patients and solitary or multinodular goiter for 3. Graves' disease proved difficult to diagnose because the thyroid gland was normal in size in two of the six patients, TRAb was positive in only two of six, and the radioiodine uptake and gradient were normal in all six; the T3-suppression test was positive in two of two patients. We conclude that the causes of subclinical hyperthyroidism are the disorders that commonly produce overt thyrotoxicosis in medical practice, Graves' disease being the most frequent. However, the tests used to diagnose overt Graves' disease often fail in the setting of subclinical hyperthyroidism, except possibly the T3-suppression test.