Adenosine stress cardiovascular magnetic resonance (CMR) has been reported to be useful for the diagnosis of coronary artery disease (CAD). Most studies use rest and stress perfusion images. The objectives of this study were to determine (Barkhausen et al. in J Magn Reson Imaging 19(6):750-757, 1) the accuracy of visual assessment and myocardial perfusion reserve index (MPRI) in the diagnosis of CAD and (Rieber et al. in Fur Heart J 27(12):1465-1471, 2) the accuracy of analysis based on rest-stress and stress images. We enrolled patients with suspected CAD and referred them for coronary angiography (CAG). All the patients underwent adenosine stress CMR before CAG. Rest and stress perfusion images were analyzed by calculation of MPRI and visual assessment separately. Visual assessment was performed separately by using rest and stress images and by using only stress images. CAG was considered the gold standard. Sensitivity, specificity, and accuracy of both methods were calculated and compared. A total of 66 patients (mean age, 61.3 +/- 11.7 years) were studied. Thirty-eight patients (57.6%) were diagnosed with CAD. The sensitivity and specificity for the diagnosis of CAD (> or =50% stenosis) were 89.5 and 78.6% for MPRI, 76.3 and 75% for stress-rest visual method, and 86.8 and 75% for stress visual method, respectively. CMR perfusion had a relatively lower accuracy in patients with left ventricular systolic dysfunction, high left ventricular mass, or presence of late gadolinium enhancement than in patients without those CMR findings. Visual assessment of stress image of CMR perfusion is accurate and comparable to MPRI for the detection of CAD.