We assessed the laboratory performance and clinical utility of a new commercial third-generation assay of thyroid-stimulating hormone (TSH), Amerlite TSH-30. The interassay CV was 6% at TSH concentrations of approximately 0.08 mIU/L, and the analytical and functional detection limits of the assay were 0.005 and 0.0125 mIU/L, respectively. Although the assay recovered approximately 96% of TSH International Reference Preparation (TSH-IRP) 80/558 added to serum samples, the endogenous TSH concentrations in basal samples were significantly lower than those found by using two other TSH assays; bias data obtained from thyroliberin stimulation tests suggested that the negative bias found with TSH-30 may be due to the heterogeneity of TSH in basal samples. TSH-30 completely discriminated hyperthyroid and hypothyroid patients from euthyroid ambulatory patients but also detected TSH (> 0.0125 mIU/L) in 3 of 46 untreated hyperthyroid patients. Compared with two second-generation assays, TSH-30 better discriminated between patients with subnormal TSH due to hyperthyroidism, thyroxine overreplacement, and nonthyroidal illness but there was still significant overlap between results for these groups.