Measurements of TSH receptor autoantibodies (TRAb) using assays based on the human monoclonal TSH receptor autoantibody M22 or bovine TSH have been compared in 136 adult patients. They suffered from Graves' disease (GD, n=62), Hashimoto's thyroiditis (HT, n=26), or non-autoimmune hyperthyroidism (NAH, n=48) and were selected on the basis of undetectable, borderline or low TRAb levels (0.6-3 IU/l) as measured by TSH based TRAb assay (Dynotest TRAKhuman from BRAHMS). The time interval between initial diagnosis of GD and TRAb determination was high and ranged from 1 month to 3.5 years (median: 2.3 years). Using the kit manufacturer's cutoff values, 53/62 (85.5%) of the selected group of GD patients were TRAb positive (>0.4 IU/l) by M22 based TRAb ELISA (Medizym TRAb clone, Medipan) and 45/62 (72.6%) were TRAb positive (>1.5 IU/l) by TSH based TRAb assay. In the HT group, 9/26 (34.6%) sera were positive in the M22 based ELISA and all but one of these 9 were positive or borderline in the TSH based assay. ROC plot analysis of the GD group using the NAH group as reference showed that at 95% specificity, the bovine TSH based TRAb assay had a sensitivity of 62.9% (cutoff for positivity=1.64 IU/l) and the M22 based TRAb ELISA a sensitivity of 90.3% (cutoff for positivity=0.32 IU/l). Overall therefore, the M22 based Medizym TRAb clone assay is more sensitive than the bovine TSH based Dynotest TRAK human assay.