The clinical usefulness of the measurement of basal TSH by an ultrasensitive assay (IRMA) versus the TRH test has been challenged in 49 children treated with L-thyroxine. They were given suppressive or replacement therapy depending on the underlying disease. An absent response of TSH to TRH could be predicted from a basal TSH value less than 0.1 mU/l in 88.8% of the cases, while only in 77.7% from a basal TSH value = 0.1 mU/l. A basal TSH value found in the range of the normal children always predicted a normal TRH test. We conclude that a sensitive TSH assay has some clinical application in monitoring L-thyroxine therapy, but can not absolutely replace the TRH test.