The objective of this study was to compare quantitative analysis of cardiac magnetic resonance (CMR) perfusion at 1.5 and 3 T against fractional flow reserve (FFR) as measured invasively. FFR is considered by many investigators to be a reliable standard to determine hemodynamically significant coronary artery stenoses. Quantitative 1.5 and 3 T CMR is capable to noninvasively determine myocardial perfusion reserve, but have not been compared against each other and validated against FFR as standard reference. Patients with suspected or known coronary artery disease (CAD) underwent CMR at at both field strengths, 1.5 and 3 T, and FFR. 34 patients were included into the study. Quantitative myocardial perfusion reserve was calculated in 544 myocardial segments at 1.5 and 3 T, respectively. FFR was measured in 109 coronary arteries. FFR ≤ 0.8 was regarded relevant. Reduced FFR (≤0.8) was found in 38 coronary arteries (19 LAD, 8 LCX and 11 RCA). Receiver operator curve analysis yielded higher area under the curve for 3 T CMR in comparison to 1.5 T CMR (0.963 vs. 0.645, p < 0.001) resulting in higher sensitivity (90.5 vs. 61.9 %) and specificity (100 vs. 76.9 %). Quantitative analysis of CMR myocardial perfusion reserve at 1.5 and 3 T is capable to detect hemodynamic significance of coronary artery stenoses. Diagnostic accuracy at 3 T is to be superior to 1.5 T.