Purpose: To test the feasibility of first-pass contrast- enhanced myocardial perfusion imaging at 3 Tesla and to evaluate the change in perfusion index between normal, remote and ischemic myocardium, we obtained perfusion index from healthy subjects and patients with coronary artery stenosis. Materials and Methods: First-pass contrast- enhanced perfusion imaging was performed on 12 patients and 32 age-matched healthy subjects in both rest and dipyridamole-induced stress states. After bolus injection of contrast agent, Gd-DTPA with dose of 0.025 mmol/kg body weight and injection time of 1.5 s, three short-axis images from apex to base of the left ventricle (LV) were acquired for 80 cardiac cycles using saturation recovery turbo FLASH sequence. The maximal upslope (Upslope) was derived from the signal-time curves of the LV cavity and myocardium to measure myocardial perfusion. Within 72 hours after cardiovascular magnetic resonance examination, patients received coronary angiography, and the results were correlated with cardiovascular magnetic resonance results. Results: Using our protocol of contrast agent administration , sufficient perfusion contrast was obtained without susceptibility-induced signal drop-out at the interface between LV cavity and the myocardium. In healthy volunteers, Upslope showed no dependence on myocardial segments or coronary territories. Upslope increased significantly from rest to stress in normal myocardium (0.09 +/- 0.03 vs. 0.16+ /- 0.05, p