MAGNETIC RESONANCE IMAGING has long been recognized as a useful tool for the non-invasive evaluation of the thoracic vasculature. Unlike computed tomography and conventional angiography, MRI is not associated with the concerns related to ionizing radiation exposure or to contrast-related nephrotoxicity. MRI is also capable of oblique image acquisition and multiplanar reformation, which aids the illustration of the thoracic vessels, inherently intertwined and complex in their arrangements. In addition, MRI using cine technique affords cardiac referenced data that enables dynamic assessment of blood flow, yielding information comparable to an echocardiogram. Gadolinium (Gd)-enhanced three-dimensional (3D) magnetic resonance angiography (MRA) is a newer technique that provides high-resolution (ie, 3D) data very quickly and is well suited for the depiction of intrathoracic vessels. Improvements in gradient technology now allow a Gd-enhanced 3D MRA to be performed during a 20–40 second breath-hold. Because it relies on T1-shortening effects of circulating Gd-chelate contrast media and not inherent flow characteristics, Gdenhanced 3D MRA can often depict pathologic vascular segments that are not adequately visualized using unenhanced flow-based MRI techniques. In addition, Gdenhanced 3D MRA provides volumetric data that can be processed for multiplanar reformation (MPR) and maximum intensity projection (MIP) viewing. In this article, the technical considerations and potential applications for Gd-enhanced 3D MRA of the systemic and pulmonary vessels within the chest will be discussed and illustrated. Traditionally, T1-weighted spin-echo and gradientecho pulse sequences have been employed for delineation of vascular pathology within the chest (1–10). The combination of T1-weighted spin-echo and gradientecho images can often provide the information necessary for the assessment of simple clinical queries such as patency of a vessel (Fig. 1) or delineation of a vascular ring (Fig. 2). These techniques, however, rely on flowing blood (ie, the movement of blood during the acquisition period) for their illustration of vascular structures. This flow dependency makes these techniques prone to flowrelated image artifacts, thereby frequently limiting their clinical utility. On spin-echo pulse sequences, vessels are characterized by their dark lumina. The black appearance of blood, also known as flow void, on spin-echo imaging occurs secondary to the wash-out of blood prior to the refocusing pulse and sampling of the echo. The washout may be incomplete if the echo time is too short, the vessel courses primarily within the imaging plane, or the blood flow is too slow. Incomplete wash-out results in the persistence of signal within the vessel lumen, which may result in the masking of underlying luminal pathology such as an intimal tear or the erroneous simulation of a vascular occlusion or thrombosis. A superior black blood effect is achieved by using preparatory pulses such as a double inversion pulse to null blood signal for more effective suppression even when the blood flow is slow. Unlike spin-echo imaging, time-of-flight (TOF) and phase contrast (PC)-MRA depict flowing blood with bright signal intensity. TOF imaging relies on the wash-in or in-flow effect of unsaturated protons; PC imaging relies, on the phase shift experienced by moving protons traveling along the gradient field (7–13). Both TOF and PC imaging provide higher intra-vascular signal-tonoise ratios than spin-echo pulse sequences. These ‘‘bright blood’’ pulse sequences (ie, TOF and PC imaging) often show intraluminal abnormalities (Fig. 1b) not clearly identified on black blood spin-echo images. However, should blood flow be slow, turbulent, or complex, vascular signal on gradient-echo images becomes unreliable. Disturbances in blood flow can often result in signal loss, which can result in the underestimation of vessel patency, overestimation of a stenosis, or even simulation of a vascular occlusion. 1Department of Radiology, University of Michigan Medical Center, Ann Arbor, Michigan 48109-0030. 2MR Research Division, Department of Radiology, Uniformed Services University, Bethesda, Maryland 20814. 3Cornell University, New York, NY 10021 Contract Grant Sponsor: Whitaker Foundation, GE. *Address reprint requests to: M.R.P., Cornell University, MRI Division, 416 East 55th Street, New York, NY 10021. Received June 29, 1999; Accepted July 13, 1999. JOURNAL OF MAGNETIC RESONANCE IMAGING 10:758–770 (1999)