Small ¹H₂O T₂* (≡ 1/R₂*) values in whole blood containing plasma gadolinium based contrast reagents (GBCRs) are widely reported. Gadoteridol (PH) relaxivities (r₂* ≡ (R₂* - R₂₀*)/[CR]: [CR] is the whole blood GBCR concentration, and R₂₀* is R₂* in the absence of GBCR) are reported from 15.07 (± 1.93) to 23.6 (± 1.8) s^-1/mM at 1.5T, and 38.64 (± 3.80) to 50.3 (± 5.4) s^-1/mM at 3.0T.^1,2 Whole blood r₂*s are much larger than plasma relaxivities and exhibit almost no r₂* variation between different GBCR chelates. In addition, the vast majority of signal dephasing is reversible by spin echo (SE; i.e., R₂ << R₂′ [≡ R₂* - R₂]). The large r₂* is likely due to GBCR exclusion from the intracellular red blood cell (RBC) space, creating a bulk magnetic susceptibility (BMS) difference between the intra- and extracellular (plasma) spaces. This produces magnetic field gradients and intracellular frequency shifts that cause signal dephasing. To better understand the molecular mechanism, we performed Monte Carlo simulations of the signal dephasing and found remarkable agreement with experimental data with no adjustable parameters.In SI units, the extracellular space magnetic susceptibility increases linearly with [CR_e] according to: $$\chi_{e}=0.31\left[CR_{e}\right]-9.1$$ while the intracellular susceptibility is constant ($$$\chi_i = - 9.19$$$, for oxygenated RBCs). If the RBC’s are approximated as ellipsoids in a uniform external magnetic field B₀, the extracellular field is inhomogeneous, but the intracellular magnetic field is uniform.³ In a general compartment c, B_c = {1 + D_c}B₀ + I_c, where D_c is the homogeneous and I_c the inhomogeneous contribution. The extracellular inhomogeneous field (I_e) can be approximated as the field surrounding a sphere of radius R: $$I_{e}=\frac{\chi_{i}-\chi_{e}}{3}\left(3\cos^{2}\alpha-1\right)\frac{R^{3}}{r^{3}}$$ where r is the positional vector from the center of the sphere, and α is the angle r makes with B₀. The extracellular homogeneous term (D_e) is equal to $$$\chi_e/3$$$. The D_i value inside an ellipsoid varies with the latter’s orientation in B₀, and is approximated as the case inside an infinite disk: $$D_{i}=\frac{\chi_e-\chi_i}{3}\left(3\cos^2\phi-1\right)+\frac{\chi_e}{3}$$ where $$$\phi$$$ is the angle the ellipsoid normal makes with B₀.Monte Carlo simulations of water molecule random walks were performed for 13,000 molecules through an ensemble of 2744 cells (spherical, R = 3.2 mm). The cell membrane water permeability coefficient (P_W) was adjusted such that the mean intracellular residence time was 10 ms (a well established value for the RBC).⁴ The diffusion coefficient was set to 1.5 μm²/ms with step duration 1 μs, thus the rms step length was 0.09 μm (Figure 1). For each step, the local Larmor frequency was calculated using either I_e or D_i, phase accumulated accordingly, and net signal calculated as the vector sum of all 13,000 spins. For spin echo simulation, the accumulated phase was reversed at intervals corresponding to the echo spacing (TE) in the previously reported in vitro experiments.² Signal data was sampled at time points corresponding to the in vitro sampling, and fitted to mono-exponential decays using non-linear least squares, yielding relaxation time constants Sim R₂* and Sim R₂. Similarly, Sim R_2e* and Sim R_2e relaxation rate constants were predicted for the extracellular inhomogeneity contribution only (i.e., the intracellular dephasing was turned off for the entirety of the walk).Figures 2-4 show the results of the full simulations (filled circles) and the extracellular contribution only (filled diamonds). The full simulation R₂ʹ agrees remarkably well with the in vitro experimental R₂ʹ (Fig. 4, open circles) for gadoteridol (PH). The full simulation accurately predicts the large, reversible dephasing. The results obtained from the extracellular dephasing predict much smaller R₂* that is mainly irreversible (i.e., the reversible component Sim R_2eʹ is near zero). Hence, intracellular dephasing due to the RBC orientation-dependent frequency differences accounts for the majority of in vitro dephasing, and is mainly reversible.With Monte Carlo simulations of signal dephasing due to intracellular frequency differences D_i and extracellular field inhomogeneities I_e, we accurately predicted transverse relaxation in oxygenated whole blood and the [CR]-dependence. Additional simulations (not shown) of the no-exchange-limit (setting P_W = 0; i.e., no trans-membrane exchange) showed that water exchange is crucial to accurately predict SE behavior. Without exchange, intracellular spins dephase only via the relatively small R₂₀ (R₂ in the absence of GBCR) in SE experiments. Thus intracellular BMS frequency differences and water exchange are essential to the GBCR-induced rapid transverse relaxation in whole blood.