The aim of this study was to evaluate the ability of the more sensitive second-generation TSH receptor (TRAb) assay to predict recurrent Graves' disease (GD) vs. remission depending on TRAb levels. 93 patients with active GD were included in the study. By using a cut-off limit of 1.0 IU/l, all 93 patients were positive for TRAb (median: 4.6 IU/l) at the time of their first visit (single point measurement in median 5.1 months after initial diagnosis). Subsequently, 33 patients went into remission and were euthyroid during follow-up (median follow-up: 21.7 months), whereas 60 patients did not go into remission or developed relapse over the following 24 months. Median TRAb levels in the group of remission were significantly (p < 0.0001) lower than TRAb values in the relapse group (2.1 compared to 8.6 IU/l). Applying ROC plot analysis to compare different TRAb thresholds, a cut-off of 10 IU/l was established. Here, the specificity for relapse was 97 % as only 1 of 29 patients with TRAb values above 10 IU/l went into remission during follow-up, whereas all other 28 patients developed a relapse (positive predictive value for relapse: 96.4 %). In contrast, TRAb values lower than 10 IU/l had no impact on the prediction of remission. In conclusion, our data clearly indicate that TRAb measurement is useful for identifying patients that will not benefit from long-term antithyroid drug treatment.