TSH-receptor autoantibodies (TRAbs) are a valuable diagnostic tool for confirming a diagnosis of Graves' disease (GD). While there is evidence that high TRAb levels are associated with relapse of GD, whether a discrimination of TRAb into stimulating (TSAb) and blocking (TBAb) autoantibodies would benefit the clinician in terms of outcome prediction remains unclear. To address this issue, we have determined TRAb, TSAb and TBAb levels in serum samples of ninety-six euthyroid patients with GD taken four weeks after antithyroid drug withdrawal (ATDT). Forty-seven patients (49 %) underwent relapse of GD within two years. Amongst those, forty-one (87 %) had been positive for TRAb and thirty-five (74 %) for TSAb after treatment. All patients except one were negative for TBAb. The correlation between TRAb and TSAb in those treated GD patients was relatively weak (r = 0.268, p < 0.001). Based on a cut-off limit of 1.5 IU/l, the positive and negative predictive values with respect to prediction of relapse were too low for any clinical relevance (TRAb: 49 % and 54 %; TSAb: 51 % and 55 %). However, when a cut-off level above 10 IU/l was used, the positive and negative predictive values increased to 83 % and 62 %. The additional measurement of TSAb or TBAb in those samples after therapy did not add additional information, even at higher decision thresholds. In conclusion, differentiation of TRAb into TSAb and TBAb is of no help in the prediction of relapse of GD in euthyroid patients at the end of ATDT, and only high TRAb levels are associated with relapse.