Vascular phantoms are used to evaluate imaging techniques such as ultrasound (US), CT, and angiography. They are expected to mimic the vasculature, surrounding tissue, and blood, and therefore must meet specific requirements on the mimicking materials, with respect to x‐ray attenuation and acoustic properties (velocity, attenuation). In the past, researchers have used a variety of vessel models, including walled (typically latex tube) and wall‐less phantoms (obtained by moulding a lumen in a block of agar). These models lacked the exact geometry of human vessels as well as pathologic features such as plaques and calcifications. To overcome these disadvantages, this paper describes a real vessel phantom for US and x‐ray studies. The phantom consists of an agar‐filled acrylic box containing a formaldehyde fixed section of a real human vessel (obtained at autopsy) cannulated onto two acrylic tubes. This phantom was evaluated by comparing the images obtained with x‐ray angiography, CT, and 3‐D B‐mode US. The images show good overall correlation based on the location of the geometrical features within the phantom, such as lumen, plaques, and calcifications. Discrepancies, artifacts, and difficulties were minor, and are discussed. The use of a real vessel, with its natural geometry and pathology, makes this phantom attractive for evaluation of imaging techniques including projection radiography, CT and US, and for extending its use to MR and US based flow studies.