Magnetic resonance imaging (MRI) is a powerful medical diagnostic technique: it can penetrate deep into tissue, provide excellent soft tissue contrast with sub-millimeter resolution, and does not employ ionizing radiation. Targeted contrast agents provide an additional layer of molecular specificity to the wealth of anatomical and functional information already attainable by MRI. However, the major challenge for molecular MR imaging is sensitivity: micromolar concentrations of Gd(III) are required to cause a detectable signal change, which makes detecting proteins by MRI a challenge. Protein-targeted MRI contrast agents are bifunctional molecules comprising a protein-targeting moiety and typically one or more gadolinium chelates for detection by MRI. The ability of the contrast agent to enhance the MR image is termed relaxivity, and it depends upon many molecular factors, including protein binding itself. As in other imaging modalities, protein binding provides the pharmacokinetic effect of concentrating the agent at the region of interest. Unique to MRI, protein binding provides the pharmacodynamic effect of increasing the relaxivity of the contrast agent, thereby increasing the MR signal. In designing new agents, optimization of both the targeting function and the relaxivity is critical. In this Account, we focus on optimization of the relaxivity of targeted agents. Relaxivity depends upon speciation, chemical structure, and dynamic processes, such as water exchange kinetics and rotational tumbling rates. We describe mechanistic studies that relate these factors to the observed relaxivities and use these findings as the basis of rational design of improved agents. In addition to traditional biochemical methods to characterize ligand-protein interactions, the presence of the metal ion enables more obscure biophysical techniques, such as relaxometry and electron nuclear double resonance, to be used to elucidate the mechanism of relaxivity differences. As a case study, we explore the mechanism of action of the serum-albumin-targeted angiography agent MS-325 and closely related compounds and show how small changes in the metal chelate can impact relaxivity. We found that, while protein binding generally improves relaxivity by slowing the tumbling rate of the complex, in some cases, the protein itself can also negatively affect hydration of the metal complex and/or inner-sphere water exchange. Drawing on these findings, we designed next-generation agents targeting albumin, fibrin, or collagen and incorporating up to four gadolinium chelates. Through judicious molecular design, we show that high-relaxivity complexes with high target affinity can be realized.