Background: Multislice computed tomography (CT) and magnetic resonance imaging (MRI) are the main candidates for noninvasive coronary angiography; however, multislice CT, unlike MRI, exposes patients to radiation and an iodinated intravenous contrast agent. Objective: To compare the diagnostic accuracy of multislice CT and MRI for noninvasive detection of clinically significant coronary stenoses (>= 50%). Design: Prospective intention-to-diagnose study. Setting: Single tertiary referral center, Berlin, Germany. Patients: 129 consecutive patients with suspected coronary artery disease. Interventions: Multislice CT and MRI were both performed within a median of 1 day before conventional coronary angiography, which served as the reference standard. Measurements: Diagnostic performance of multislice CT and MRI. Results: 129 patients completed the study. Altogether, 108 patients with 430 vessels could be examined with both multislice CT and MRI and were used for analysis. In the per-patient analysis, the sensitivity of multislice CT (92% [95% CI, 82% to 96%]) was significantly higher than that of MRI (74% [CI, 61% to 83%]; P = 0.013). The sensitivity for detecting clinically significant stenoses was 82% for multislice CT and 54% for MRI (P < 0.001). Specificity and negative predictive value of multislice CT and MRI in the per-vessel analysis were 90% versus 87% (P = 0.73) and 95% versus 90% (P = 0.032), respectively. The effective radiation dose used with multislice CT (mean, 12.3 mSv [SD, 1.4]) in a consecutive subgroup of 73 patients was not significantly different from that used with diagnostic cardiac catheterization (11.4 mSv [SD, 4.8]) (P = 0.169). Most patients (74%) indicated that they would prefer multislice CT for future diagnostic imaging (P < 0.001). Limitations: This was a single-center study with 129 patients. Conclusions: In patients referred for conventional coronary angiography, multislice CT compares favorably with MRI for noninvasive detection of coronary stenoses.