Intraplaque hemorrhage, enlarging lipid cores, and their proximity to the flow lumen are important determinants of carotid plaque rupture and neurological complications. We developed an image-analysis method for B-mode ultrasound, pixel distribution analysis (PDA), for preprocedural identification of these high-risk features in carotid plaques. This technique may improve selection of patients for carotid endarterectomy and carotid artery stenting. Forty-two patients with high-grade carotid stenosis in 45 arteries, 18 symptomatic and 27 asymptomatic, underwent preoperative ultrasound. Intraplaque hemorrhage, lipid, fibromuscular tissue, calcium, lipid core area, and distance from the flow lumen were quantified using pixel intensities of tissues in control subjects. These findings were contrasted between symptomatic and asymptomatic plaques and correlated with histology. Inter- and intraobserver variabilities were determined for this technique. Pixel intensities of control tissues were discrete and significantly different from each other (median: blood 0, lipid 27, muscle 45.5, fibrous tissue 204, and calcium 245). There was more intraplaque hemorrhage (p<0.001) and lipid (p=0.002) but less calcium (p<0.001) within symptomatic plaques. Lipid cores were larger (p=0.005) and their distance from the flow lumen was lower (p=0.01) in symptomatic plaques. Intraplaque hemorrhage, lipid, fibromuscular tissue, calcium, lipid core size, and distance from flow lumen measured by PDA correlated with histology. No significant inter- or intraobserver variabilities were observed in these measurements. PDA accurately identified more intraplaque hemorrhage and lipid, less calcium, and larger lipid cores located closer to the flow lumen in symptomatic patients with carotid stenosis. These data indicate that PDA may be used to identify high-risk carotid atherosclerotic plaques and thereby improve the selection of patients requiring treatment.