Thyrotropin receptor (TSH-R) gene expression can be positively or negatively regulated by TSH and stimulating TSH-R antibodies (TSAbs) in immortalized thyroid cell lines such as rat FRTL-5 cells. However, regulation is less clear in other mammalian cells including cultures of human thyroid cells. Additionally, it has been suggested, based on FRTL-5 cell data, that TSH-R gene negative regulation by TSH or TSAbs might be lost in Graves' disease. The present study evaluated TSH-R gene transcript levels in thyroids from patients with Graves' disease to correlate in vivo data with in vitro observations or hypotheses. TSH-R mRNA levels were characterized in a total of 66 human thyroid glands with particular concern to levels in Graves' patients. Results were related to clinical parameters, transcript levels of thyroglobulin (TG), and thyroid peroxidase (TPO), as well as transcript levels of thyroid transcription factor 1 (TTF-1) which regulates the expression of all three genes and paired box-gene 8 (Pax-8) which regulates TG and TPO gene expression. Northern blot analyses showed that TSH-R expression was significantly increased, 2.2-fold, in Graves' thyroids (p = 0.0098, n = 35) by comparison to normals (n = 6). TSH-R mRNA levels were decreased to 30% and 7% of normal levels in Hashimoto's thyroids (p = 0.0281, n = 5) and anaplastic carcinomas (p = 0.0033, n = 6), respectively. No significant changes were seen in endemic goiters (n = 8) and in thyroid autonomy (n = 6). TSH-R RNA levels were higher, 3.6-fold, in thyroids of a subgroup of Graves' patients that had not been pretreated with iodide before surgery (n = 10) by comparison to thyroids from those that had been treated before surgery, 1.7-fold (n = 25). TSH-R antibodies exhibited a nonsignificant tendency toward a negative correlation. All other clinical or endocrine parameters showed no clear relation to TSH-R mRNA levels. Pax-8 and TTF-1 transcripts were detectable in normal thyroids; however, Pax-8 expression was increased in Graves' thyroids (3.8-fold), whereas TTF-1 expression was only minimally changed in all thyroids investigated. Changes of the two did not correlate. Pax-8 expression correlated with TG and TPO expression (in all cases, p = 0.0001); TTF-1, despite its minimal change, still correlated with TG (p = 0.0471) but not with TPO expression (p = 0.0984). TTF-1, again despite its minimal changes, correlated positively with TSH-R gene expression (p = 0.0251); however, surprisingly, Pax-8, which does not regulate TSH-R gene expression, correlated even better with TSH-R transcript levels (p = 0.0001). We conclude that augmentation of TSH-R expression levels, and thus potential ligand binding sites, may indicate an important regulatory principle in the pathogenesis of autoimmune hyperthyroidism in vivo: the responsiveness of the TSH-R to TSH and TSAb induced negative regulation is lost. This increase of TSH-R expression levels is not due to an ongoing transcriptional activation of the TTF-1 gene. Pax-8, though positively correlated with TSH-R RNA levels, cannot be the factor either, because Pax-8 does not upregulate TSH-R expression. This predicts that other factors involved in TSH-R induced negative regulation are abnormal and must be searched for and evaluated.