The purpose of this study was to prospectively determine the accuracy of low-dose computed tomography coronary angiography (CTCA) for the diagnosis of functionally relevant coronary artery disease (CAD) using cardiac magnetic resonance (CMR) as a standard of reference. Forty-one consecutive patients (age 64 +/- 10 years) underwent k-space and time broad-use linear acquisition speed-up technique accelerated CMR (1.5 T) and dual-source CTCA using prospective electrocardiography gating within 1 day. CTCA lesions were analysed and diameter stenoses of more than 50% and more than 75% were compared with CMR findings taken as the reference standard for assessing the functional relevance of CAD. CMR revealed perfusion defects in 21/41 patients (51%). A total of 569 coronary segments were analysed with low-dose CTCA. The image quality of low-dose CTCA was diagnostic in 566/569 segments (99.5%) in 39/41 patients (95%). Low-dose CTCA revealed stenoses of more than 50% in 58/123 coronary arteries (47.2%) in 24/41 patients (59%) and more than 75% stenoses in 46/123 coronary arteries (37.4%) in 23/41 patients (56%). Using a greater than 50% diameter stenosis, low-dose CTCA yielded the following per artery sensitivity, specificity, positive and negative predictive values, and accuracy for the detection of perfusion defects: 89%, 79%, 72%, 92% and 83%, respectively. Low-dose CTCA is reliable for ruling out functionally relevant CAD, but is a poor predictor of myocardial ischaemia.