The diagnosis of transient ischaemic attack (TIA) is based largely on the patient?s symptom recall and clinical judgement. This decision-making process is highly subjective and the inter-observer reliability of TIA diagnosis is at best moderate, even among neurologists. The aim of this study is to examine the presenting features and final diagnoses of referrals to a TIA clinic and to evaluate characteristics that favoured the diagnosis of TIA over other TIA ?mimics?. Consecutive new referrals to a tertiary metropolitan hospital TIA clinic over a 9 month period were examined. Characteristics between TIA and non-TIA diagnoses were compared and analysed. Eighty-two patients were recruited. Eighteen (22 ) were given a final diagnosis of TIA or stroke. Major alternative diagnoses included migraine (n = 17, 21 ), presyncope/syncope (n = 13, 16 ) and anxiety (n = 7, 9 ). Four (5 ) patients had unclassifiable symptoms with no clear final diagnosis. Mean age was 67 ? a standard deviation of 17 years and patients diagnosed with TIA/stroke were on average older than those with non-TIA diagnoses (77 ? 10 versus 64 ? 17 years, p = 0.003). A diagnosis of TIA/stroke was favoured in the presence of moderate to severe weakness (p = 0.032), dysphasia (p = 0.037) or dysarthria (p = 0.005). Unclassifiable symptoms (for example, palpitations, confusion, headache) were reported in 27 patients (33 ) and their presence favoured non-TIA diagnoses (p = 0.0003). TIA constituted a minority of the referrals to our clinic. Accurate clinical diagnosis of TIA facilitates early stroke prevention and avoids unnecessary investigations and prescriptions. Attempts to improve diagnostic accuracy of TIA should target improving the education and awareness of frontline medical practitioners.