PURPOSE: To prospectively evaluate the feasibility and diagnostic accuracy of high spatial resolution myocardial perfusion imaging during high dose dobutamine/atropine stress magnetic resonance (DSMR) for the detection of coronary artery disease (CAD). METHODS AND RESULTS: DSMR-wall motion was combined with perfusion imaging (DSMR-perfusion) in 78 patients prior to clinically indicated invasive coronary angiography. For DSMR-perfusion an in-plane spatial resolution of 1.5×1.5mm(2) was attained by using 8×k-space and time sensitivity encoding (k-t SENSE). Image quality and extent of artifacts during perfusion imaging were evaluated. Wall motion and perfusion data were interpreted sequentially. Significant CAD (stenosis≥70%) was present in 52 patients and involved 86 coronary territories. One patient did not reach target heart rate despite maximum infusion of dobutamine/atropine. Two studies (3%) were non-diagnostic due k-t SENSE related artifacts resulting from insufficient breathhold capability. Overall image quality was good. Dark-rim artifacts were limited to the endocardial border at a mean width of 1.8mm. The addition of DSMR-perfusion to DSMR-wall motion data improved sensitivity for the detection of CAD (92% vs. 81%, P=0.03) and accurate determination of disease extent (85% vs. 66% of territories, P<0.001). There were no significant differences between DSMR-perfusion and DSRM-wall motion regarding overall specificity (83% vs. 87%, P=1) and accuracy (89% vs. 83%, P=0.13). CONCLUSION: High spatial resolution DSMR-perfusion imaging at maximum stress level was feasible, improved sensitivity over DSMR-wall motion for the detection of CAD and allowed an accurate determination of disease extent. Specificity of DSMR-perfusion with k-t SENSE improved compared to prior studies using lower spatial resolution.