First-pass MR myocardial perfusion measurements require a well-defined left ventricular (LV) blood pool input function. We used a peripheral intravenous (i.v.) injection of a gadolinium (Gd) chelate to obtain a well-characterized LV time-intensity curve. Using a strongly T1-weighted subsecond MR sequence, we performed cardiac MR imaging after administering an IV bolus injection of one of three different doses of the Gd chelate: a standard dose (0.1 mmol/kg, group I, n = 8); a low dose with two bolus volumes (0.01 mmol/kg, 1/10e bolus volume, group II, n = 7, and 0.01 mmol/kg diluted in saline, same bolus volume as group I, group III, n = 3); and an intermediate dose (0.05 mmol/kg, group IV, n = 5). Unlike in group I (high dose), in groups II and III (low dose), the LV curve had a well-defined first peak, followed by a downslope and a recirculation peak. With the intermediate dose (group IV), a saturation effect still remained on the LV curve. The signal intensity (SI) enhancement of the myocardium was respectively 580 +/- 77% at 0.1 mmol/kg, 362 +/- 95% at 0.05 mmol/kg, and at 0.01 mmol/kg, it was 184 +/- 33% in group II and 272 +/- 8% in group III. In conclusion, with subsecond T1-weighted MR imaging and a low dose of Gd chelate (i.e., 0.01 mmol/kg), the LV input function is a well-defined first step for MR perfusion modeling.