Gadolinium contrast agents (such as gadopentetate dimeglumine, Gd-DTPA) combined with dynamic MR studies are increasingly used to measure physiological properties of biological systems. In a typical application, the concentration of Gd-DTPA is traced in vivo following intravenous administration. Since Gd shortens the relaxation time of nearby protons, the MR signal can be measured in tissue regions, converted to the absolute concentration, and the resulting time course analyzed, yielding functional parameters.The advantage of dynamic MRI with Gd-DTPA over iodinated contrast agents imaged with X-rays and scintigraphic studies includes a lack of radiation exposure and a low rate of adverse reaction. While the relationship between tracer dose and the accuracy of the measurements is well understood in nuclear medicine and CT imaging applications, little work has been devoted to analyze this relationship in dynamic MRI.We describe a method for computing the concentration c of the tracer from T 1 -weighted images that are characterized by large signal nonuniformities (such as torso phasedarray or surface coil images). For a wide range of values, c can be uniquely derived from measured MR signal S, the precontrast signal intensity SЈ, and the precontrast longitudinal relaxation time TЈ. We show how the fixed errors in measured variables propagate through computed parameters. Finally, based on realistic arterial and tissue response functions measured in both normal subjects and subjects with established renal insufficiency, we perform Monte Carlo simulations to optimize the dose of Gd-DTPA for two quantitative dynamic MR studies: 1) the measurement of cardiac output and 2) the measurement of glomerular filtration rate.