BackgroundTSH receptor antibody (TRAb) is considered the gold standard diagnostic test for the autoimmunity of Graves’ disease (GD), which is commonly diagnosed clinically.AimTo evaluate the true positive (sensitivity) and true negative (specificity) rates of clinical diagnosis of GD or non-GD hyperthyroidism compared to the TRAb test.SettingUniversity teaching hospital in North West England.ParticipantsPatients in the Endocrinology service who had a TRAb measurement between December 2009 and October 2015.MethodsElectronic patient records were studied retrospectively for a pre-TRAb clinical diagnosis of GD or non-GD hyperthyroidism. We examined descriptive statistics and binary classification tests; Fisher exact test was used to analyse contingency tables.ResultsWe identified 316 patients with a mean age of 45 (range, 17–89) years; 247 (78%) were women. Compared to the TRAb result, clinical diagnosis had a sensitivity of 88%, specificity 66%, positive predictive value 72%, negative predictive value 84%, false negative rate 12%, false positive rate 34%, positive likelihood ratio 2.6 and negative likelihood ratio 0.2 (P < 0.0001).ConclusionsClinicians were liable to both over- and under-diagnose GD. The TRAb test can help reduce the number of incorrect or unknown diagnoses in the initial clinical assessment of patients presenting with hyperthyroidism.