PURPOSE: To evaluate the hemodynamic causes of signal inhomogeneity at the carotid bulb that might be misinterpreted as pathologic signal defect in carotid contrast enhanced MRA(CEMRA). MATERIALS AND METHODS: Both carotid CEMRA and fast digital subtraction angiography(DSA) were conducted on 15 patients (28 carotid arteries) and arterial phase CEMRA images were compared with fast DSA images of the same patients. A 1.5T MR imager was used. The Turbo-FLASH sequence emplayed was TR/TE/FA= 3 .2m s / 1.3m s / 35 degree. For experimental study, we utilized handmade silicon phantoms of the tortuous carotid bifurcation; these might be expected to clearly demonstrate turbulent flow at the carotid bulb. In a closed circulatory system, both CEMRA and fast DSA involved the use of these phantoms. RESULTS: During CEMRA, inhomogeneous signals of varying degrees were found at the carotid bulb in 12/28 carotid arteries. When compared with sequential DSA images, incomplete mixing of contrast agent due to turbulent flow at the carotid bulb might be responsible for this inhomogeneity. This hypothesis was reinforced by successfully reproducing signal defects at the carotid bulb from the experimental CEMRA study using carotid phantoms that showed marked turbulent flow in the same area during DSA. CONCLUSION: Incomplete mixing of contrast agent caused by turbulent flow at the carotid bulb might be responsible for the signal inhomogeneity seen on carotid CEMRA.