The development of highly sensitive immunometric assays for thyroid-stimulating hormone (TSH) has provided increased understanding of thyroid hormone regulation but, paradoxically, has contributed to a kaleidoscopic complexity of thyroid function test variability in hospitalized patients with nonthyroidal illness (NTI). In primary hypothyroidism, an elevated TSH is the most sensitive chemical index available, although early cases may show a hyperresponse of TSH to thyrotropin-releasing hormone (TRH) stimulation when the TSH is still within the normal range. The ability of the new TSH assays to discriminate between normal and low levels now allows the diagnosis of thyrotoxicosis to be confirmed by a suppressed TSH in the presence of elevated serum thyroxine (T4) and/or triiodothyronine (T3). The TRH stimulation test is virtually obsolete for the diagnosis of thyrotoxicosis but remains of much interest in the investigation of psychiatric syndromes. Approximately 25% of patients with depression have a blunted TSH response (a rise of less than 5 microU/mL) that differs from thyrotoxicosis, wherein the TSH response is suppressed under 1 microU/mL. The cause of the blunted TSH is uncertain but is not due to hyperthyroidism. In contrast, close to 15% may have a TSH hyperresponse to TRH and/or elevated antithyroid antibodies. Thyroid hormone treatment may benefit the depression in some of these cases. In the sick thyroid state of nonthyroidal illness, a low T3 level is the initial manifestation. In more severe cases, the T4 also falls, the free T4 level in this situation is variable, both normal and low levels being reported from different laboratories. A diagnosis of hypothyroidism requiring treatment with thyroid hormone therapy is unlikely unless there is a concomitant lowfree T4 and elevated TSH in a patient who is not in the process of recovery. In acute psychiatric admissions, there is a high frequency of hyperthyroxinemia. The TSH in these cases is generally either normal or high, suggesting central activation of the hypothalamic-pituitary-thyroid axis. In most instances, the thyroid function tests normalize within 2 weeks, and treatment directed toward the thyroid gland is not indicated. Suppressed TSH levels, usually associated with a normal free T4, has also been described in such patients. Finally, various medications utilized in psychiatric practice have diverse effects on thyroid function and can cause diagnostic difficulty. These include lithium, phenytoin sodium, and carbamazepine, and their effects are reviewed.