Nuclear magnetic resonance measurements provide useful insight into pore-scale properties of porous media. One influence affecting the measurement is inhomogeneity in the static background magnetic field (). This inhomogeneity can significantly alter the relaxation signal and potentially obscure pore-scale information. To improve the understanding of this effect on the free-induction decay (FID), a measurement still commonly used in practice, a novel methodology is developed to estimate the statistical distribution of . A suite of preparatory pulse sequences is developed to encode information about the field in the initial amplitude and phase of the FID following each sequence and an inversion is employed to predict the statistical distribution of . Knowledge of the distribution is then used to correct for the impact of inhomogeneity on the FID measurement; this is essential for improving the usefulness of FID measurements for the estimation of pore-scale properties. Results are presented for both numerical and laboratory studies verifying the feasibility of the developed methodology in a controlled laboratory environment, and demonstrating that knowledge of the statistical distribution of is sufficient to estimate the impact of inhomogeneity on the FID in cases where inhomogeneity causes less than an order of magnitude decrease in the relaxation times governing the FID.