Because a significant number of all patients seen by cardiologists have had coronary bypass surgery, a relatively noninvasive method of assessing coronary bypass graft patency would be very helpful. Ultrafast computed tomography, by virtue of its rapid data acquisition time and reasonable spatial resolution, may be useful in this regard. To determine the sensitivity, specificity and predictive accuracy of this imaging modality as compared with cardiac catheterization, a multicenter study was undertaken. There were two parts to the study. Part I involved the evaluation of 179 grafts in 74 patients studied in the five participating centers between March 1985 and August 1986. Twenty-nine percent of these graft studies were found to be technically inadequate and were excluded before patency determinations began. The remaining group of 127 bypass grafts in 62 patients had studies adequate for interpretation. Fifty-one grafts werl to the left anterior descending coronary artery or a diagonal branch, 37 to branches of the left circumflex artery and 28 to the right coronary artery or a posterior descending vessel; in addition, there were 11 internal mammary artery bypass grafts primarily into the left anterior descending or diagonal artery distribution. The sensitivity of detecting angiographically open grafts was 93.4%, the specificity of detecting angiographically closed grafts 88.9% and the predictive accuracy was 92.1%. A subsequent study (Part 2) was performed 9 months later to assess the ability to carry out technically adequate examinations. Of the 138 consecutive graft examinations (50 patients) included in this part of the study, 94.2% of the examinations were found to be technically adequate. From these data it is concluded that 1) Ultrafast computed tomography is a very useful, minimally invasive technique for evaluating coronary artery bypass graft patency, and 2) technically adequate studies can be effectively performed in the majority of patients with bypass grafts.