Current immunometric TSH assays guarantee a functional sensitivity of at least 0.10 mU/l. The 95% confidence interval of TSH reference values is 0.45-4.12 mU/l; it becomes slightly broader above the age of 60 years (0.4-5.9 mU/l for 70-79 years). Interference by heterophilic antibodies in the TSH assay may give rise to spuriously elevated TSH values. The TSH assay is a very accurate diagnostic test for screening thyroid function. A normal TSH indicates euthyroidism with an accuracy of > 99%; only the rare patient with central hypothyroidism may be overlooked, but the history usually provides sufficient clues to suspect hypothalamic or pituitary disease. In case of an abnormal TSH value, a FT4 assay should be done. Each of the six possible outcomes (decreased or increased TSH combined with a decreased, normal or increased FT4 value) points to a specific diagnosis with an accuracy of over 90%. In the follow-up of treatment of thyroid function disorders, TSH target values are slightly different from the normal reference values. In primary hypothyroidism one aims at TSH values between 0.25 and 2.0 mU/l, but should avoid TSH values < or = 0.10 mU/l. In differentiated thyroid cancer, following complete thyroid ablation the target value is < or = 0.10 mU/l. A decreased TSH value in the course of treatment for Graves' hyperthyroidism does not necessarily indicate thyrotoxicosis, but may reflect persisting levels of TSH receptor stimulating antibodies.