A new immunochemiluminometric TSH assay (ICMA) was shown to offer improved analytical (+2 SD of zero) and functional (20% interassay coefficient of variation) sensitivity [0.003 vs 0.045 +/- 0.005 (+/- SE; range, 0.01-0.07); 0.018 vs. 0.23 +/- 0.02 (range, 0.10-0.35, mU/L); analytical vs. functional sensitivity limit for the ICMA vs. 10 other TSH immunometric assays, respectively]. The ICMA was used to study the physiological relationship between serum TSH and free T4 [as reflected by free T4 index (FT4I)] values at both steady state and 14 days after acute pharmacological T4 administration (3 mg oral T4 load plus 0.3 mg daily). At steady state, an inverse log/linear relationship was found between serum TSH and FT4I values (log TSH = 2.56 - 0.022 FT4I; r = 0.84; P less than 0.001). Ten to 14 days after acute T4 suppression in 5 euthyroid subjects, serum TSH/FT4I levels had plateaued after decreasing in parallel to the slope of the steady state relationship, suggesting that the degree of T4 suppression of TSH can be predicted from an individual's pituitary TSH/free T4 set-point and the magnitude of the serum T4 elevation achieved. Ambulatory and hospitalized patient sera, previously identified as having low (less than 0.1 mU/L) TSH levels by a less sensitive assay, were restudied by the TSH ICMA. Normal TSH values ranged from 0.39-4.6 mU/L, whereas the majority of hyperthyroid patients [52 of 54 (96% ambulatory) and 22 of 23 (96%, hospitalized)] had undetectable (less than 0.005 mU/L), basal TSH levels and absent TRH stimulated TSH responses. In contrast, most (32 of 37; 86%) of hospitalized nonhyperthyroid patients with low (less than 0.1 mU/L) TSH values due to nonthyroidal illness or glucocorticoid treatment had detectable (greater than 0.01 mU/L) basal and TRH stimulated TSH levels. The positive relationship between basal and TRH-stimulated TSH levels was shown to extend down to the detectability limit of the assay (0.005 mU/L), which further supported the authenticity of the subnormal TSH ICMA measurements. The new TSH ICMA is considered to represent the first of a third generation of clinical TSH assays, since it has a functional (interassay) sensitivity that is 2 orders of magnitude greater than that of typical first generation TSH RIAs and 1 order of magnitude greater than current second generation TSH immunometric methods. Such third generation TSH assays will facilitate both the optimization of T4 therapy as well as the diagnosis of hyperthyroidism in hospitalized patients with nonthyroidal illness.