Background: There is paucity of data about arterial wall characteristics of the smallest and largest caliber cerebral vessels. Determining the relationship between the lumen and the wall might shed new insights into cerebral artery remodeling. Objective: To test the hypotheses that arteries with larger luminal diameters have a thinner wall and that arteries with the smallest lumina have thicker walls. Methods: Cross-sectional segments from large arteries (N=1392) were obtained from the circle of Willis in 196 autopsied brains (mean age 55 ± 17 yrs, 39% with hypertension, 15% with diabetes and 20% with dyslipidemia). Lumen diameter, stenosis percentage, and thicknesses of intima, media, and adventitia were calculated in digital microphotography after staining. Atheromas and internal elastic lamina (IEL) disruption were rated visually. Arteries were categorized into the top 5% (“dilated”) and bottom 5% (“narrowed”) of the luminal diameters, as well as an intermediate category (90% of sample as reference). We used logistic regression to obtain the odds of association (OR, 95% CI) after adjusting for demographic and vascular variables. Results: Narrowed arteries were more frequently found in men (OR 2.7, 95%CI 1.3-5.9) and with dyslipidemia (4.2, 1.6-11.1) while dilated arteries were more frequently found in women (5.6, 2.2-14.0), in smokers (2.6, 1.0-6.5) and those with prior MI (7.7, 1.2-48.7). Narrowed arteries were more likely to have atheromas (20.8, 4.8-90.3), greater luminal stenosis (per %, 1.1, 1.1-1.2), thicker vessel walls (1.3, 1.2-1.4), but thinner medias (0.9, 0.8-1.0). Conversely, larger arteries exhibited less IEL disruption (0.3, 0.1-0.9), atheromas (0.34, 0.1-0.9) and stenosis (0.8, 0.8-0.9), their walls were thinner (0.8, 0.8-0.9) but the media was thicker (1.1, 1.1-1.2). Conclusions: Narrowed cerebral arteries were more likely to have atheromas while dilated arteries had thinner walls and were more frequent in subjects with prior MI. These findings suggest that both extremes of the arterial spectrum might be differentially related with vascular disease, underscoring the need to revisit whether standard preventive measures for vascular disease are equally effective in patients harboring such disparate arterial phenotypes.