Aim of our study was to evaluate the increasing sensitivity within three generations of thyroglobulin (Tg) assays, which were available during the past decade, and its clinical impact for patients with differentiated thyroid carcinoma. METHODS: Determination of Tg using the IRMA introduced in 1989 (Dynotest Tg, Henning Berlin, Berlin; assay A) and 1994 (Selco Tg, Medipan Diagnostica, Selchow; assay B), as well as the IEMA available recently (Medizym Tg Rem, Medipan Diagnostica, Selchow; assay C). RESULTS: We found a close correlation between the measurable Tg values of assay A and B (r=0.985; p<0.001) as well as assay B and C (r=0.978; p<0.001). Assay B (lowest detection limit: 0.3 ng/ml) was more than twice as sensitive as assay A and did not show quite as many disturbances of recovery (in 0.5% versus 4% of our patients). Due to its strict calibration to the European reference preparation CRM 457, Tg values determined by assay C were in the mean 1.9-fold higher than by assay B. Thus, with its functional sensitivity of 0.03 ng/ml assay C is nearly 20-fold more sensitive than assay B. Whereas the proportion of measurable Tg values was only 22% in a selected group of patients (criterion of inclusion: Tg in assay B< or =1 ng/ml with TSH-suppressive conditions; n=317 serum samples from 103 patients), it was 68% in assay C, with good intraindividual reproducibility of these values in the course. CONCLUSION: The ultrasensitive assay C is especially suitable for the follow-up of treated thyroid cancer patients being considered as cured, and may shorten the time interval until the detection of a recurrence markedly: the gain of time calculated from the Tg courses in patients with a gradually progressive tumor relapse ranged from 5 to 15 months.