Current clinical applications of dynamic contrast-enhanced magnetic resonance imaging are based on the extravasation of relatively small contrast agents (SCAs). The high extravasation rate of SCAs is considered a disadvantage, as it requires fast imaging protocols at the expense of image quality, spatial resolution, or anatomic coverage. Medium-sized contrast agents (MCAs) leak more slowly into tissue and would therefore allow longer dynamic acquisition times, enabling improved image quality. Here the influence of molecular size on the reliability of pharmacokinetic parameters, i.e. the volume transfer constant Ktrans, extracellular extravascular fraction ve, and plasma fraction vp, was investigated. Computer simulations, with in vivo measured arterial input functions, were performed to determine the bias and variance of the pharmacokinetic parameters as a function of contrast agent size, sampling frequency, noise level, and acquisition time. Better reliability of all parameters was obtained for the MCA compared to the SCA. To obtain similar variance in Ktrans, fs for the SCA (28 min-1) needed to be 20 times faster than for the MCA (1.3 min-1). For MCAs, vp estimation was reliable at fs >3.5 min-1, while SCAs require fs > 60 min-1 was required. For the MCA, variance decreased for increasing Ta. Reliable estimation was realized for Ta > 5 min for Ktrans and Ta > 13 min for ve.For MCAs lower sampling frequencies can be used compared to SCAs and better reliability was obtained for Ktrans, ve, and vp. Compared to SCAs, longer acquisition times were required for MCAs to obtain maximum identifiability of ve. With MCAs, reliable estimation of vp can be obtained at much lower frequencies than with SCAs. In conclusion, MCAs provide superior identifiability of pharmacokinetic parameters compared to SCAs.