Introduction: Emergency department management of transient ischaemic attack varies from admission for all to outpatient referral. We studied the short-term outcomes of transient ischaemic attack managed with an agreed protocol. Predictors of stroke can be different for Asians and non-Asians. ABCD2 as initial triage of transient ischaemic attack is debatable. The predictive ability of ABCD2 score was studied as well. Methods: This was a prospective observational study with consecutive subject recruitment in Emergency department. All transient ischaemic attacks were admitted, hard and e-records of Emergency department, transient ischaemic attack clinic, Medical and Neurosurgical department and general follow-ups in Hospital Authority hospitals were studied up to 1 year. Stroke-day was measured from symptom-onset to time-of-stroke. Results: In 18-month period, 124 patients were recruited. The median onset-to-door time was 3.5 h. All computed tomography brain positive findings, except one subdural haematoma, were ischaemic in origin. Six strokes, all disabling, recurred within 90 days, three on day 1–3, two died in 6 months. The stroke risks at 2, 7, 90 days and 1 year were 1.61%, 3.23%, 4.84% and 4.84%, respectively. No significant trend was observed in stroke risk across ABCD2 scores ( p = 0.783) with area under the curve of 0.537 (95% confidence interval = 0.380–0.694; p = 0.762). The short-term stroke risk was associated with atrial fibrillation ( p = 0.002). The median Emergency Medicine ward length of stay was 1.33 days. Conclusion: In our Emergency department–based management, the short-term stroke risk of transient ischaemic attack is low, and the predictivity of ABCD2 score in risk stratification cannot be validated. Stroke recurrences were associated with atrial fibrillation. A low ABCD2 could be falsely reassuring. As half of the strokes recurred very early, we recommend admission in the hyperacute phase.