Recent increases in the sensitivity of methods used to measure TSH, especially third generation assays, have enabled separation of partial from complete pituitary suppression in patients with thyrotoxicosis. We have investigated the use of a sensitive chemiluminescent enzymeimmunoassay in the differential diagnosis of thyrotoxicosis. Serum TSH concentrations were determined by chemiluminescent enzymeimmunoassay in patients with various types of overt and subclinical thyrotoxicosis. The assay was highly sensitive with an analytical sensitivity of 0.0016 mU/l. Among 45 hyperthyroid patients with untreated Graves' disease, 37 (82.2%) showed undetectably low levels (< 0.002 mU/l). Serum TSH in the remaining 8 patients (17.8%) was 0.003-0.005 mU/l. In contrast, TSH was undetectably low in only 5 (20.0%) of 20 patients with painless thyroiditis and in 2 (12.5%) of 16 patients with subacute thyroiditis. Eleven (55.0%) with painless thyroiditis and 12 (75.0%) with subacute thyroiditis had TSH values greater than 0.005 mU/l (0.006-0.032 and 0.006-0.228 mU/l, respectively; normal range 0.5-5.0 mU/l). Serum TSH levels were subnormal in 10 of 12 patients with euthyroid ophthalmic Graves' disease, including 4 with undetectably low levels. Among 11 patients with an autonomously functioning thyroid nodule 6, including 1 with normal free T4 and free T3 and 2 with normal free T3, showed TSH values less than 0.002 mU/l. No significant correlation was observed between serum free T4 or free T3 and TSH concentrations in thyrotoxic patients. Together with the incomplete suppression of TSH observed in those with destructive thyroiditis, this suggests that the grade of TSH suppression was influenced by the duration of illness at the time of blood sampling. The third-generation TSH assay is useful for the differential diagnosis of various types of thyrotoxicosis, especially between Graves' disease and destructive thyroiditis.