Coronary angiographic studies performed with 16-channel multidetector computer tomographic scanners have demonstrated accurate detection of coronary vessel stenosis but are limited by a significant number of non-evaluable segments. To date, only single-center experience with multidetector computer tomography has been reported. We performed a prospective, blinded study at 2 institutions to determine the feasibility and diagnostic accuracy of coronary angiography using 40-channel multidetector computer tomography with multi-segment reconstruction for the detection of obstructive coronary artery disease (CAD). Multidetector computer tomographic studies were performed in 85 patients who were referred for invasive coronary angiography with clinically suspected CAD. Datasets were analyzed by blinded, independent review. Of 1,145 segments that were suitable for analysis as determined by angiography, 1,045 (91.3%) were evaluable on multidetector computer tomography. Segment-based sensitivity, specificity, and positive and negative predictive values for detecting > or =50% luminal stenoses were 86%, 97%, 75%, and 97%, respectively. The area under the receiver-operating characteristic curves for the detection of > or =50% angiographic stenosis by multidetector computer tomography was 0.94. In a patient-based analysis, the sensitivity, specificity, and positive and negative predictive values for detecting subjects with > or =1 segment with > or =50% stenosis were 98%, 93%, 94% and 93%, respectively. In conclusion, coronary angiography using 40-channel multidetector computer tomography with multi-segment reconstruction accurately detects coronary segments and patients with obstructive CAD, with a small number of non-evaluable cases.