Angiographic procedures were carried out on 36% of 1,328 patients suspected of having transient ischemic attacks (TIA). Among six participating centers, this ranged from 13% to 82%. This large difference might be related to the number of patients considered good surgical candidates and differences in the use of screening noninvasive diagnostic techniques. Arch studies, using catheter techniques, were performed most often. Although 13% of the patients had transient complications, permanent neurological deficits occurred in only 0.65%. Angiographic lesions were best correlated to clinical symptoms in those patients thought to definitely have carotid artery system TIA but were commonly seen in all other groups. Thus, clinical correlation was poor.