The roles of B-mode real-time ultrasonic imaging (USI) and carotid angiography (CAG) in deciding on endarterectomy were analyzed. When greater than 50% stenosis and/or complicated ulceration were predicted by both techniques, endarterectomy was undertaken in 21/25; by only CAG in 5/8; and by only USI in 10/21. In vessels with adequate surgical data, stenosis was accurately predicted by CAG in 27/32, USI in 20/32, and ulceration by both CAG and USI in 21/29. USI results were sometimes used to justify endarterectomy despite unimpressive CAG. CAG is a better predictor of stenosis, whereas both are limited in predicting ulceration.