The purpose of this study was to compare the performance of 2D vs. 3D time-of-flight (TOF) methods in imaging the normal pulmonary arteries with commercially available 1.0 T equipment. The study was conducted in 20 volunteers and 7 patients with suspected pulmonary embolism (PE). To reduce artifacts caused by cardiac and respiratory motion, MR images were acquired in volunteers using two-dimensional (2D), gradient-recalled echo (GRE), breath-hold techniques, and three-dimensional (3D) acquisitions. Sagittal thin (6-MM) segmented k-space 2D sections obtained with cardiac gating during systole (turboFLASH, TR/TE9/6 ms, 14 segments of 9 lines) and incremented flip-angles (TONE), and 50-mm 3D volume TONE acquisitions with 32 partitions (FISP, TR/TE34/10ms) were successively performed. In the second phase of the study, patients were examined only with the 3D technique. Images of volunteers were qualitatively and quantitatively analyzed. S/N ratios were statistically compared by means of the paired-sample Wilcoxon ranked-signed test, a value of p < .05 being significant. In volunteers, 3D acquisitions displayed significantly more segment-order pulmonary arteries on average than did 2d acquisitions displayed significantly more segment-order pulmonary arteries on average than did 2D acquisitions (2.95 +/- 0.64 vs. 2.2 +/- 0.85, respectively; p < .01). Moreover, the signal intensity of arteries within the lungs was less homogeneous in the 2D than in the 3D technique, with a signal intensity ratio between peripheral and proximal arteries of 63% +/- 7% and 73% +/- 2%, respectively (p < .05). In patients, no erroneous diagnoses were obtained using the 3D technique. 3D images of normal lungs provide MR angiograms of better quality than do 2D images, and require less contribution from subjects because they are performed in free breathing. Ongoing improvements in MR sequences and further studies are now necessary to assess the value of 3D TONE MRA in the diagnosis of PE.