Purpose: Magnetization transfer saturation ((Formula presented.)) is a useful marker to probe tissue macromolecular content and myelination in the brain. The increased (Formula presented.) -inhomogeneity at (Formula presented.) T and significantly larger saturation pulse flip angles which are often used for postmortem studies exceed the limits where previous (Formula presented.) (Formula presented.) correction methods are applicable. Here, we develop a calibration-based correction model and procedure, and validate and evaluate it in postmortem 7T data of whole chimpanzee brains. Theory: The (Formula presented.) dependence of (Formula presented.) was investigated by varying the off-resonance saturation pulse flip angle. For the range of saturation pulse flip angles applied in typical experiments on postmortem tissue, the dependence was close to linear. A linear model with a single calibration constant (Formula presented.) is proposed to correct bias in (Formula presented.) by mapping it to the reference value of the saturation pulse flip angle. Methods: (Formula presented.) was estimated voxel-wise in five postmortem chimpanzee brains. “Individual-based global parameters” were obtained by calculating the mean (Formula presented.) within individual specimen brains and “group-based global parameters” by calculating the means of the individual-based global parameters across the five brains. Results: The linear calibration model described the data well, though (Formula presented.) was not entirely independent of the underlying tissue and (Formula presented.). Individual-based correction parameters and a group-based global correction parameter ((Formula presented.)) led to visible, quantifiable reductions of (Formula presented.) -biases in high-resolution (Formula presented.) maps. Conclusion: The presented model and calibration approach effectively corrects for (Formula presented.) inhomogeneities in postmortem 7T data.