Background: The majority of TIAs last from 2 to 15 minutes, although some may be of long duration. Objective: We examined factors related to the duration of TIAs to identify the relationship to clinical characteristics. Methods: We performed brain imaging studies as well as angiographic and cardiac examinations in 81 consecutive patients (64 men and 17 women, age 65.8 +/- 9.9 years) with carotid TLAs. We evaluated risk factors (hypertension, diabetes mellitus, hyperlipidemia, alcohol consumption, and smoking), potential cardiac sources of emboli, and arterial stenosis greater than or equal to 50% in diameter in the carotid or middle cerebral arteries. Recent infarcts were assessed with CT or MRI. We correlated duration of symptoms with clinical data. Results: The presence of emboligenic cardiac or arterial diseases was significantly related to the duration of symptoms. With sensitivity-specificity curve analysis for detecting such diseases, the duration of symptoms could be divided into short-duration TLAs (<60 minutes, n = 41) or long-duration TIAs (greater than or equal to 60 minutes, n = 40). Patients with long-duration TIAs had emboligenic cardiac or arterial diseases more frequently than those with short-duration TIAs (86% versus 46%, p < 0.001). Recent infarcts were also more frequent in patients with long-duration Tills than they were in patients with short-duration TLAs (45% versus 21%, p < 0.05). Conclusions: Short-duration and long-duration TIAs can be separated based on symptom duration of <1 hour or greater than or equal to 1 hour. Patients with long-duration TLAs should be examined more closely for the presence of cardiac and arterial diseases than those with short-duration TIAs.