To evaluate the sensitivity of basal TSH concentrations as determined by an "ultrasensitive" IRMA-assay (RIA-gnost h-TSH-monoclonal, Behring) versus a "negative" TRH test (defined as an increment of TSH ≥0.2 mU/l 20 min after administration of 400 μg TRH iv) in the diagnosis of hyperthyroidism we examined 193 consecutive patients from our thyroid outpatient clinic: 34 patients displayed hyperthyroidism (total T4: 184.4±26.0 μmol/l, effective thyroxine index: 1.25±0.08), whereas 12 had isolated T3-hyperthyroidism (total T3: 3.47±0.48 nmol/l). Employing the producer's definition of subnormal ("suppressed") bTSH concentrations (≤0.1 mU/l), only 19 (41.3%) hyperthyroid patients would have been detected; on the other hand, one euthyroid patient would have been recognized false positively as hyperthyroid. Using the TRH test as criterion led to the correct diagnosis in 42 (sensitivity: 91.3%) hyperthyroid patients, whereas two had low bTSH concentrations (≤0.5 mU/l), but a normal TSH response to TRH (>2.0 mU/l). Raising the threshold concentration to 0.2 and, subsequently, to 0.4 mU TSH/l increased the number of correct results to 38 (sensitivity: 82.6%) and 43 (93.5%), respectively. This was associated with a concomitant decrease in specificity in the diagnosis of hyperthyroidism from 93.7 (0.1 mU/l) to 27.9% (0.4 mU/l). In conclusion, despite ultrasensitive methods for estimation of low TSH concentrations, the TRH test remains an irreplaceable tool for the correct diagnosis of hyperthyroidism.