The aims of this study were to evaluate the diagnostic value of orthogonal magnetic resonance angiography (MRA) and to compare the results of MRA with those of digital subtraction venography (DSV) in thoracic venous diseases. Ten normal volunteers were evaluated using two-dimensional time-of-flight MRA in three orthogonal planes to determine the image quality of each venous segment. Twelve consecutive patients suspected of having thoracic venous disease were studied with both MRA and DSV. In the normal subjects, the plane perpendicular to the target vein provided the most consistent visualization. Using three orthogonal MRA images, a diagnostic-quality image was obtained in 175 (83%) of 210 venous segments in normal volunteers. In patients with thoracic venous obstructive disease, MRA was more effective than DSV in detecting total (84 vs 54), patent (56 vs 36), stenotic (13 vs 10), and obstructive (15 vs 8) venous segments, poststenotic or postobstructive veins (15 vs 10), thrombosis of the internal jugular vein (7 vs 2), intraluminal thrombus (5 vs 3), and azygos veins (12 vs 2). Using venous segments visible on DVS (n = 54) as the standard, the sensitivity and specificity of MRA were 94% and 100%, respectively, in detecting venous patency, and 100% and 98% in detecting complete venous obstruction. In the shoulder region, the sensitivity and specificity of MRA were 93% and 100%, respectively, in detecting venous patency, and 100% and 97% in detecting venous obstruction. We conclude that MRA with three orthogonal planes can provide relatively complete and reliable venous mapping, without the need for contrast medium.