Despite the earlier accepted notion that CEA should be delayed 4 to 6 weeks after a stroke, current evidence suggests that CEA may be performed safely earlier than this in most patients who have mild to moderate deficits once symptoms stabilize. The gray areas, however, remain gray, as outlined. Crescendo TIAs are urgent cases in the authors' practice; others advocate a more moderate delayed approach in such cases. Almost everyone agrees that propagating intraluminal thrombus is treated best with a moderate delayed approach that allows the thrombus to resolve first with anticoagulants. Acute carotid occlusion must be assessed on an individual basis: cases that occlude under observation should be explored immediately; cases that come from the field with profound deficits have dismal outcomes, but even here surgery may be effective in salvaging a small group of good functional survivors, and the natural history without surgery is atrocious. Surgery for stroke in evolution is associated with higher morbidity and mortality rates; selected patients may benefit from emergency surgery. A final thought is that for patients who have routine TIA or small stroke, with minimal imaging evidence of infarction or mass effect, a stable deficit, and a normal level of consciousness, there is no reason to empirically delay carotid reconstruction, and patients are served best by a fast-track approach to surgical treatment.