We evaluated interobserver and intraobserver variability in the arteriographic assessment of the carotid bifurcation. Two neuroradiologists evaluated area stenosis and lumen surface morphology for 60 biplane carotid arteriograms in two reading sessions. Absolute interobserver differences for area stenosis averaged 8.8%, and intraobserver differences were statistically significant for arteries of at least 50% stenosis. Analysis of observations for lumen surface morphology revealed statistically significant interobserver differences, averaging 29.5% disagreement. Intraobserver differences were not statistically significant. Observer variability for arteriography should be considered in the process of evaluating patients for carotid endarterectomy.