The appropriateness of use and accuracy of age, blood pressure, clinical features and duration of symptoms (ABCD) scoring by non-stroke specialists while risk-stratifying patients with suspected transient ischaemic attack (TIA) are unknown. We reviewed all available ABCD data from referrals to a specialist neurovascular clinic. ABCD scoring was defined as 'appropriate' in this study if an experienced vascular neurologist subsequently confirmed a clinical diagnosis of possible, probable or definite TIA, and 'inappropriate' if the patient had an alternative diagnosis or stroke. Interobserver agreement between the referring physician and the neurologist was calculated. One hundred and four patients had completed ABCD referral proformas available for analysis. Forty-five (43%) were deemed appropriate, and 59 (57%) inappropriate. In the entire dataset, the neurologist agreed with the referring physician's total ABCD score in only 42% of cases [κ = 0.28]. The two most unreliable components of the scoring system were clinical features [κ = 0.51], and duration of symptoms [κ = 0.48]. ABCD scoring by non-stroke specialists is frequently inappropriate and inaccurate in routine clinical practice, emphasising the importance of urgent specialist assessment of suspected TIA patients.