A method for measuring transverse relaxation rate, R2*, corrected for the presence of macroscopic magnetic field inhomogeneities with a magnetic resonance imaging (MRI) system is provided. The method accounts for additional signal decay that occurs as a result of macroscopic variations in the main magnetic field, B0, of the MRI system, and also mitigates susceptibility-based errors and introduction of increased noise in the R2* measurements. Image data are acquired by sampling multiple different echo signals occurring at respectively different echo times. A B0 field inhomogeneity map is estimated by fitting the acquired image data to an initial signal model. Using the estimated field map, a revised signal model that accounts for signal from multiple different chemical species and for signal decay resulting from macroscopic variations in the B0 field is formed. Corrected R2* values for the different chemical species are then estimated by fitting the acquired image data to the revised signal model.