We compared the conspicuity of acute pulmonary emboli with contrast-enhanced spiral computed tomography (CT) and two- and three-dimensional time-of-flight magnetic resonance (MR) techniques. Seven dogs who received experimental pulmonary emboli and one control were imaged with spiral CT and with 2-D (FMPVAS and FASTCARD) and 3-D time-of-flight MR. Blinded, independent, prospective evaluations of the CT and MR images by two MR radiologists and two chest radiologists were then compared to the location of the emboli as determined by subsequent pathologic evaluation of the excised lungs. Embolus/blood contrast-to-noise ratios (CNRs) were calculated on both MR and CT images for pulmonary emboli that could be identified. Fifty emboli ranging from 1.0 to 5.5 mm (mean, 2.7, +/- 0.14 SEM) in diameter and from 3.0 to 60 mm (mean, 28.1 +/- 1.9 SEM) in length were found in the seven embolized dogs on pathologic examination. Three of the four radiologists identified more thrombi on CT images than they did on their best MR pulse sequence (FASTCARD) and with greater confidence. The fourth radiologist identified an equal percentage of clot on CT and FASTCARD images with confidence slightly greater on FASTCARD MR than on spiral CT. Mean CNR for the best MR technique was 43.4 (+/- 3.9 SEM) and for CT was 20.7 (+/- 1.3 SEM). In general, pulmonary emboli were detected more accurately on contrast-enhanced spiral CT than on MR. This occurred although the embolus/blood CNR was higher on MR than on CT. Better pulmonary embolus conspicuity on CT images was attributed to better spatial resolution and fewer artifacts on CT than on MR. One MR radiologist performed equally well with both spiral CT and FASTCARD techniques, suggesting that experience may be a factor in performance.