Spoiled gradient echo pulse (SPGRE) sequences are commonly used in dynamic contrast-enhanced MRI (DCE-MRI) studies to measure the contrast agent concentration in a tissue of interest over time. However, due to improper tuning of the SPGRE parameters, concentration uncertainty can be very high, even at high signal-to-noise ratio in the MR measurement. In this work, an optimization procedure is proposed for selecting the optimal value of the SPGRE-flip angle FA(opt), given the expected concentration range. The optimization condition ensures that every concentration in the assumed range has the lowest possible uncertainty. By decoupling the R(1)- and R*(2)-effects caused by the presence of the contrast agent, a contour plot has been generated from which FA(opt) can be read off for any study design. Investigation of ten recent DCE-MRI studies showed that improper flip angle selection unnecessarily increases the concentration uncertainty, up to 742% and 72% on average for the typical physiological concentration ranges of 0-2 mM in tumour tissue and 0-10 mM in blood, respectively. Simulations show that the reduced noise levels on the concentration curves, observed at the optimal flip angle, effectively increase the precision of the kinetic parameters estimates (up to 82% for K(trans), 82% for nu(e) and 92% for nu(p) in the case of an individually measured arterial input function (AIF), up to 53% for K(trans), 59% for nu(e) and 67% for nu(p) in the case of a standard AIF). In vivo experiments confirm the potential of flip angle optimization to increase the reproducibility of the kinetic parameter estimates.